Gnome BODY (important!)

You're pushing this Luck angle pretty hard, and I have to say it's starting to wear pretty thin. 
 
I'd never suggest replacing Combat Luck with Luck, for a few reasons. 
A - People want and need consistency in their RDEF.  One botched roll meaning days of bedrest is simply unacceptable for the tone of many (I want to use "heroic" as an adjective with different meaning) high-flying low-lethality games.  One botched roll meaning you lose a giant chunk of BODY and STUN if you don't have other RDEF makes combat too swingy.  Telling a player their Combat Luck can't be reliable will be met with characters suddenly deciding to wear armored costumes or get force-belts or mutate to be bulletproof. 
B - Rolling dice every time you get attacked pushes the rolls-per-attack even higher, and it's already at least three for most games.  HERO gets slow enough with newbies and won't-memorize-rulesbies already.  It'd work for an experienced group, maybe. 
C - Combat Luck doesn't have to be actual Luck despite the name.  Reference the first paragraph of its description in FRED. 
D - Luck itself is a clusterexpletive because it includes no useful guidelines on when to actually roll or what each success on the dice actually means.  It's great at one table and trash at another. 
 
Combat Luck fills a rather definite niche that is highly distinct from the one filled by Luck, and the two should never be merged.  I will gladly say that Combat Luck is badly named though. 

You're pushing this Luck angle pretty hard, and I have to say it's starting to wear pretty thin. 
 
I'd never suggest replacing Combat Luck with Luck, for a few reasons. 
A - People want and need consistency in their RDEF.  One botched roll meaning days of bedrest is simply unacceptable for the tone of many (I want to use "heroic" as an adjective with different meaning) high-flying low-lethality games.  One botched roll meaning you lose a giant chunk of BODY and STUN if you don't have other RDEF makes combat too swingy.  Telling a player their Combat Luck can't be reliable will be met with characters suddenly deciding to wear armored costumes or get force-belts or mutate to be bulletproof. 
B - Rolling dice every time you get attacked pushes the rolls-per-attack even higher, and it's already at least three for most games.  HERO gets slow enough with newbies and won't-memorize-rulesbies already.  It'd work for an experienced group, maybe. 
C - Combat Luck doesn't have to be actual Luck despite the name.  Reference the first paragraph of its description in FRED. 
D - Luck itself is a clusterexpletive because it includes no useful guidelines on when to actually roll or what each success on the dice actually means.  It's great at one table and trash at another. 
 
Combat Luck fills a rather definite niche that is highly distinct from the one filled by Luck, and the two should never be merged.  I will gladly say that Combat Luck is badly named though. 

Really bad hilt ergonomics? 

This really really depends on the GM.  Some are overly sensitive about players questioning rulings and/or house rules, and will react poorly.  Some will take it in stride but bluntly refuse to consider a change.  Some will hear out the player's arguments and make an impartial decision.  Some will bend over backwards. 
The correct approach depends on the GM, and only MechaniCat can know what his GM is like. 
Since I don't know MechaniCat's GM, my advice earlier to the thread was for MechaniCat to talk things over with his GM about this and make sure to phrase it as "I'm not sure if my concept fits" instead of "Your rules are crippling my character". 

Builds characters which way? 
 
I don't obsessively powergame for maximum impact within a set of constraints (at least if I'm submitting the character for a game, theoretical optimization isn't practical optimization) if that's what you're asking. 
I do spend all my points, take all my disads, and ask the GM for a rough idea of how combat capable a character should be.  My GM doesn't do hard caps, so I can't say I buy up to them but I can say I don't leave gaping weaknesses or seriously underbuy anything. 
 
I've had a few games where someone brought a complete mechanical trainwreck to a 4-man game where the GM was running a module intended for 4 PCs.  The result of the party effectively being down a man was generally a ruined game, anger, and accusations of incompetence and/or powergaming.  This has utterly soured me on the idea of deliberate under-performance.  A character who's too bad can be as disruptive as a character who's too good. 

Let d be the distribution of defenses in the campaign. 
 
Let x(d) be the mean damage of a successful attack by character A's attack.  For a Blast or similar power, this would be (X/5)*3.5-DEF where DEF is whatever defense applies against this attack.  So long as (X/5)*3.5-DEF >> 0, this is a good approximation.
Let var(x(d)) be the variance in x(d), which in turn is just the variance in DEF.
 
Let y1(d), y2(d), y3(d) be the mean damages of B's attacks.  Let y(d) be the maximum of y1(d), y2(d), y3(d).  Let var(y(d)) be the variance in y(d), which will be lower than var(x(d)).  Computing this is a bit harder, but still doable. 
 
Let W be the value given to mean damage-per-phase and V be the value, likely negative, given to variance. 
The relative value of A's powers and B's powers can be expressed as W*(mean(x(d))-mean(y(d))) + V*(var(x(d))-var(y(d))).  If this has a positive result, character A is optimal for this distribution of d and choice of W and V.  If this has a negative result, character B is optimal. 
This is a bit unsatisfying to me since it's admittedly just a fancy way to say "depends on the campaign and how much you want stable damage output", but it's a fancy mathematical way that could be made into a computer program to very decisively answer the question for a given d X Y W and V. 

Let d be the distribution of defenses in the campaign. 
 
Let x(d) be the mean damage of a successful attack by character A's attack.  For a Blast or similar power, this would be (X/5)*3.5-DEF where DEF is whatever defense applies against this attack.  So long as (X/5)*3.5-DEF >> 0, this is a good approximation.
Let var(x(d)) be the variance in x(d), which in turn is just the variance in DEF.
 
Let y1(d), y2(d), y3(d) be the mean damages of B's attacks.  Let y(d) be the maximum of y1(d), y2(d), y3(d).  Let var(y(d)) be the variance in y(d), which will be lower than var(x(d)).  Computing this is a bit harder, but still doable. 
 
Let W be the value given to mean damage-per-phase and V be the value, likely negative, given to variance. 
The relative value of A's powers and B's powers can be expressed as W*(mean(x(d))-mean(y(d))) + V*(var(x(d))-var(y(d))).  If this has a positive result, character A is optimal for this distribution of d and choice of W and V.  If this has a negative result, character B is optimal. 
This is a bit unsatisfying to me since it's admittedly just a fancy way to say "depends on the campaign and how much you want stable damage output", but it's a fancy mathematical way that could be made into a computer program to very decisively answer the question for a given d X Y W and V. 

I think I got sloppy about clarity somewhere in the reply chain.  Let me be more formal. 
Assume: Defenses are generally bought relative to attacks such that a NND attack against a "normal" opponent and a blast of the same AP have roughly equal STUN output.  Based on the guidelines in FRED and the characters I have seen made by my group, posted on the forum, and in published materials, I believe this to be a reasonable assumption for superheroic play. 
Let character A have a pair of N point attack powers that are able to be used in a multiple power attack.  Let character B have a single N point reserve multipower with N points spent on ultra slots containing attacks.  Omit consideration of other powers A or B may have. 
Observation: A can multiple power attack, B cannot.  Observation: B can flexibly change the defense they are targeting, A cannot. 
Conclusion: A will have greater raw output, but be highly susceptible to variation caused by their target's characteristics.  B will experience the inverse, with a stable but lower output. 

I think I got sloppy about clarity somewhere in the reply chain.  Let me be more formal. 
Assume: Defenses are generally bought relative to attacks such that a NND attack against a "normal" opponent and a blast of the same AP have roughly equal STUN output.  Based on the guidelines in FRED and the characters I have seen made by my group, posted on the forum, and in published materials, I believe this to be a reasonable assumption for superheroic play. 
Let character A have a pair of N point attack powers that are able to be used in a multiple power attack.  Let character B have a single N point reserve multipower with N points spent on ultra slots containing attacks.  Omit consideration of other powers A or B may have. 
Observation: A can multiple power attack, B cannot.  Observation: B can flexibly change the defense they are targeting, A cannot. 
Conclusion: A will have greater raw output, but be highly susceptible to variation caused by their target's characteristics.  B will experience the inverse, with a stable but lower output. 

I think I got sloppy about clarity somewhere in the reply chain.  Let me be more formal. 
Assume: Defenses are generally bought relative to attacks such that a NND attack against a "normal" opponent and a blast of the same AP have roughly equal STUN output.  Based on the guidelines in FRED and the characters I have seen made by my group, posted on the forum, and in published materials, I believe this to be a reasonable assumption for superheroic play. 
Let character A have a pair of N point attack powers that are able to be used in a multiple power attack.  Let character B have a single N point reserve multipower with N points spent on ultra slots containing attacks.  Omit consideration of other powers A or B may have. 
Observation: A can multiple power attack, B cannot.  Observation: B can flexibly change the defense they are targeting, A cannot. 
Conclusion: A will have greater raw output, but be highly susceptible to variation caused by their target's characteristics.  B will experience the inverse, with a stable but lower output. 

@Killer Shrike: It's worth noting that 60AP of attack against 20 DEF, 12 RDEF, no other defenses results in roughly 20 STUN for most forms of offense.  All having a massive spread lets you do is ignore uneven increases in defense and exploit weaknesses. 
 
@Toxxus: If A buys two attacks while B buys an attack MP, A can multiple attack and throw twice the offensive weight compared to B.  That's a pretty huge advantage. 

For me, it's that GURPS has one building block and a lot of exceptions where HERO has many building blocks but few exceptions. 
I can look at a HERO character sheet and tell what everything does without looking anything up, since I know the base powers and the advantages and the limitations.  It took a while to get there, but now I've "learned HERO". 
I can't look at a GURPS character sheet and tell what things do without looking them up, because each one-line entry needs a book-and-page lookup to find what the unique text of that ability or flaw is.  Unless I memorize every GURPS book, I've never "learned GURPS". 

That sounds to me like they're always broken.  Either you get more than you pay for or you no longer have what you paid for.  Both are bad. 

In addition to everything Doc Democracy mentioned, consider the inverse situation of a socially skilled NPC attempting to change the opinion of a PC.  Without resolution mechanics, it is dangerously easy to fall into the horrible patterns of "Tray Tor approaches you and proposes a mutually beneficial deal, speaking convincingly of a plan he hatched to -" "I say no, and go hit on the waitress." "But he's very convincing abou-" "My character is unconvinced.  I'm gonna go hit on the waitress." "No, your character is convinced and agrees to help!" "No, my character is unconvinced and hitting on the waitress!". 
For much the same reason as violent combat needs rules to determine who can do and withstand what, social combat needs rules to indicate when someone is fooled, convinced, agitated, or otherwise forced to act against their better interests.  Otherwise you just get 'nuh-uh' 'uh-uh' 'nuh-uh' 'uh-uh'. 

In addition to everything Doc Democracy mentioned, consider the inverse situation of a socially skilled NPC attempting to change the opinion of a PC.  Without resolution mechanics, it is dangerously easy to fall into the horrible patterns of "Tray Tor approaches you and proposes a mutually beneficial deal, speaking convincingly of a plan he hatched to -" "I say no, and go hit on the waitress." "But he's very convincing abou-" "My character is unconvinced.  I'm gonna go hit on the waitress." "No, your character is convinced and agrees to help!" "No, my character is unconvinced and hitting on the waitress!". 
For much the same reason as violent combat needs rules to determine who can do and withstand what, social combat needs rules to indicate when someone is fooled, convinced, agitated, or otherwise forced to act against their better interests.  Otherwise you just get 'nuh-uh' 'uh-uh' 'nuh-uh' 'uh-uh'. 

For me, it's that GURPS has one building block and a lot of exceptions where HERO has many building blocks but few exceptions. 
I can look at a HERO character sheet and tell what everything does without looking anything up, since I know the base powers and the advantages and the limitations.  It took a while to get there, but now I've "learned HERO". 
I can't look at a GURPS character sheet and tell what things do without looking them up, because each one-line entry needs a book-and-page lookup to find what the unique text of that ability or flaw is.  Unless I memorize every GURPS book, I've never "learned GURPS". 

If we assume that everyone has a smartphone, then everyone's on equal footing.  But if Long Wave wants to buy the power to tap into the cell network with his mind, he shouldn't be paying full price for something he'd already have for free.  If we assume everyone has access to a 2d6 HKA or RKA from mundane equipment, then everyone's on equal footing.  But if Gonedolf wants to buy the power to shoot arrows with his mind, he shouldn't be paying full price for something he'd already have for free  So on and so forth. 
But at the same time, not having the Focus limitation (and Real Weapon, and etc etc) means the power-version is superior to the equipment-version (barring worse limitations on the power version, of course). 
 
The solution is to make players only pay the difference.  If cell phones are free and worth 12 AP with -1 in Limitations, then a cell phone power should cost 6 fewer points.  If 6 DCs of KA with -1 in Limitations can be obtained from equipment for free, then a PC who buys a KA should be paying only to remove those limitations or add DCs. 

You are incorrect because your opinions are opinions and my opinions are facts.  As proof, I am of the opinion that my opinions are facts, therefore it is fact that my opinions are facts. 
Furthermore I am of the opinion that 6e's removal of the Lack of Weakness power was terrible since no longer can the GM respond to "I use Find Weakness on him" with "FOOL, DOCTOR DEFENESTRATION HAS NO WEAKNESS!".  Therefore 5e is objectively and unarguably superior to 6e in all ways. 

It's the Resistant advantage being applied to existing PD and ED. 

In a normal game, the challenges and dangers involve Rayzer Blade stabbing you with a laser gun, getting punched by Obligatory Cold Pun's massive ice-clad fists, or having your secrets stolen by Suckerberg's mind-sharing powers. 
FRED quite wisely proscribes using social skills on PCs for the reasons you mentioned. 
 
But to me, "political/intrigue game" suggests that the challenges and dangers should involve cunning courtesans wooing you so they can manipulate you into acting on their behalf (or so they can stab you when you're unarmored), courtly rivals flinging veiled insults to goad you into brash and reckless actions (possibly ones that end in their bodyguards punching you), or a masterful statesman leading you into a trap of words that leads to you accidentally admitting something unpleasant (like your secrets). 
Put more generally, I would take "political/intrigue game" to mean one where the proscriptions are, by necessity, lifted.  You could, I admit, run a game where the PCs are always "on the attack" in social scenes and everything boils down to them initiating the die-rolls and them being cunning and clever and manipulative.  But that seems to me like it'd gut the ability of NPCs to be cunning and clever and manipulative and place the PCs in this bizarre state where only certain highly specific lines of social attack as determined by their disads are possible. 
 
Overall, to me, "political/intrigue game" requires players to opt into the idea of having less agency.  Communicating that fact to prospective players would be critical. 

In a normal game, the challenges and dangers involve Rayzer Blade stabbing you with a laser gun, getting punched by Obligatory Cold Pun's massive ice-clad fists, or having your secrets stolen by Suckerberg's mind-sharing powers. 
FRED quite wisely proscribes using social skills on PCs for the reasons you mentioned. 
 
But to me, "political/intrigue game" suggests that the challenges and dangers should involve cunning courtesans wooing you so they can manipulate you into acting on their behalf (or so they can stab you when you're unarmored), courtly rivals flinging veiled insults to goad you into brash and reckless actions (possibly ones that end in their bodyguards punching you), or a masterful statesman leading you into a trap of words that leads to you accidentally admitting something unpleasant (like your secrets). 
Put more generally, I would take "political/intrigue game" to mean one where the proscriptions are, by necessity, lifted.  You could, I admit, run a game where the PCs are always "on the attack" in social scenes and everything boils down to them initiating the die-rolls and them being cunning and clever and manipulative.  But that seems to me like it'd gut the ability of NPCs to be cunning and clever and manipulative and place the PCs in this bizarre state where only certain highly specific lines of social attack as determined by their disads are possible. 
 
Overall, to me, "political/intrigue game" requires players to opt into the idea of having less agency.  Communicating that fact to prospective players would be critical. 

In addition to everything Doc Democracy mentioned, consider the inverse situation of a socially skilled NPC attempting to change the opinion of a PC.  Without resolution mechanics, it is dangerously easy to fall into the horrible patterns of "Tray Tor approaches you and proposes a mutually beneficial deal, speaking convincingly of a plan he hatched to -" "I say no, and go hit on the waitress." "But he's very convincing abou-" "My character is unconvinced.  I'm gonna go hit on the waitress." "No, your character is convinced and agrees to help!" "No, my character is unconvinced and hitting on the waitress!". 
For much the same reason as violent combat needs rules to determine who can do and withstand what, social combat needs rules to indicate when someone is fooled, convinced, agitated, or otherwise forced to act against their better interests.  Otherwise you just get 'nuh-uh' 'uh-uh' 'nuh-uh' 'uh-uh'. 

It's the Resistant advantage being applied to existing PD and ED. 

I disagree with both those, actually. 
Longevity should be free unless NND Longevity attacks are reasonably frequent (IE, going to come up).  Any benefits from living long, such as knowledge, should be bought normally with the "very old" SFX.  Even if NNDs are in the equation, 1 point is about all I'd support it costing. 
The +2 to Concealment should come with a mandatory purchase of Concealment so it actually works.  Or just be replaced with a mandatory purchase of Concealment. 

Hit Locations: A quick-and-dirty fix that comes to mind would be shift hit locations to "general multiplier" and "specific multiplier".  The general multiplier is always no higher than 1 and applies to all damage, the specific multiplier is always no lower than 1 and applies to STUN for normal attacks and BODY for KAs.  Thus, a hit to the head with a normal attack would deal more STUN but no extra BODY, but the reverse for a KA.  A hit to the foot would do less STUN and BODY always. 
 
Increased Stun Multiple: I'd suggest reworking this into something more along the lines of attack 8d6 + 2d6 (Adds STUN only -1).  Add extra dice of STUN or BODY as appropriate to skew the desired outputs. 
I'm aware that STUN only is normally -0, but that's because it fully nonlethalizes an attack.  Partial nonlethalization is nowhere near as useful. 
 
Penetrating: Why not just divide penetrating BODY by 3?