But as a logical proposition, the OP's question was "are feats about A, or are they just about B?" Under the latter clause, specifically, if they're just about B, then that precludes them also being about A (or, indeed, C, D, or E). At which point, I label that clause "false", and look at A.
There is an edge care here, and it is that edge case I'm arguing for and that is when A = B. In this case, something can be both about A and just about B because we can restate the phrase as "are feats about A, or are they just about A?" Since being "just about A" implies being "about A" we can answer "Yes" or maybe more precisely, "False contradiction", since the original poster seemed to be implying A is mutually exclusive with B (B = !A), which is false if (B = A). If the OP wasn't implying they are contradictory (inclusive or), then A = true and B = true implies A || B = true.
It remains to show that A = B.
By Feat I mean, "An optional benefit that can be selected and gained by multiple classes as a part of character advancement", to distinguish it from Traits (which are often listed with feats but lack the characteristic of feats in being gained in character advancement) and class abilities which are listed with feats but are actually only obtainable within the frame work of the progression of a single class (these can be variously named, but for now lets use the term 'Mutations' for the fun of it). This is a rather particular definition but I think it is a good one. I'm having to make this particular definition to avoid some edge cases (that I previously alluded too), namely a feat that isn't a benefit ("paying to suck"?). However, since the feat structure includes the word 'benefit' right there in the text, we can I think safely assume that 'benefit' is or was the intention.
Proving Feats "just make the character better" is now trivial. By definition that's what a benefit does. All feats by definition "just make the character better".
But, proving that Feats are customization is nearly as easy. Customization means, "To make or alter to individual or personal specifications". Since feats are optional, specific, and alter the characteristics of the character, they are customization by definition. Not every customization is a feat, and not everything that "just makes the character better" is a feat, but the entire circle I've just labeled Feat fits within the space of "customization" and "makes everything better". Moreover, "makes something better" is also entirely within the space of customization (again, by definition) so it is possible then to say that Feats are A and also just B, because we can find no counter case where the Feat is "just B" but not A.
We can imagine feat replacements and things that look like feats but aren't that are customizations but don't "just make the character better", but these edge cases ('Traits', 'Disadvantages', 'Mutations', etc.) are really not feats but choices that replace feats (or which can sometimes be replaced by feats) or which allow for the choice of additional feats. They are not feats as I've formally defined them, and in some cases ('Disadvantages') they wouldn't be recognized as being true feats.
I would agree with this. But then, I'm probably biased - even from before 3e was released, I was convinced Prestige Classes were a terrible idea. Far better, IMO, to make either the base classes themselves more customisable or provide a far greater range of feats. Either way, PrCs become unnecessary.
We seem to be in agreement on this.
(Actually, my preferred approach would be to move those feats that give new powers, such as Power Attack, into a separate category (talents)
Power granting feats have always been problematic for me, and most of them are in my opinion kludges - late recognition of a shortcoming in the system that was most easily addressed by having feat. I generally use what I call "The Playground Test". If the power is one which we can imagine two elementary students using on one another in their playground scuffle, it probably shouldn't be a feat. So for example, I would object to a Feat that allowed you to make a trip attack; conversely, I would not object to a feat that allowed you to perform a trip attack better. It's never been completely clear to me that Power Attack doesn't fail the play ground test, if it represents something like a "haymaker" or other telescoped attack. If that is the case, Power Attack should become a generic combat option, and the feat "Improved Power Attack" would then be a feat and not what you call 'a talent'.
The problem is that almost all talents potentially fail this test. It isn't clear that for example, the "High School" test or "Graduate Student" test isn't equally applicable as an evaluator. If we say that 'a talent' represents the result of more particular sorts of training, it's not clear that it isn't simply some more difficult application of skill and ability that should be available to everyone of sufficient skill and ability and that the best implementation is not move it in to the generic skill or combat system and then making a feat that represents the ability to perform this relatively early and well (by this theory, all feats are simply variants on Skill Focus). I've come close to deciding this should be a hard rule several times, and I'm certainly much closer to that in my implementation than stock 3.X and just about all 3rd party implementations. There are however a few cases where I've just felt it more interesting to silo the power within a feat, typically by making it an extension of or alternate use of an existing skill or where the feat grants a miniature version of what is normally a class ability without the prerequisite of taking the class (say the ability to cast cantrips or orisons).