Handled by background and skill selection. A rogue subclass of warrior could give the class expertise or additional skill selections if you wanted.
OK. I'll buy that. If needed, spend a feat to gain X skills.
Taken to its logical conclusion, though, that train of thought gets to character points for build (one class with the ability to select/buy features). That then goes to character points for advancement. I like those systems just fine, but it wouldn't be D&D. The class/level system is the heart of what makes D&D. It adds enough structure to avoid analysis paralysis for casual/new players, but (the heart of changes since late 2E) adds enough options to avoid boredom in advanced players.
One of the things that killed 3E was the feat/class/thing bloat. By the end, it would have been less complex to use Hero or GURPs. There's a balance.
You have to have enough options, without having too many. That's probably a range, but it's still a problem of scarcity -- only so many slot to fill, so use them wisely. If people feel some meaningful options aren't being met (warlord, psion, artificer), and we're anywhere near that sweet spot, it makes sense to ask what resources are being spent poorly. I don't know what "the number" is, but my guess is somewhere in the 10-15 range*, with sub-classes multiplying that by virtue of turning one question into two. Right now, we've got twelve classes, which is right in the middle of that range.
Personally, I'd happily trade the Sorcerer for a good Psion or the Barbarian for an Artificer. That does go back to your point of subjectivity. But, that's what discussion is for: seeing if there's common thoughts on a topic.
* This number is very subjective. I picked a range that "feels right" for mass consumption.