One of the authors of the third edition, Monte Cook, has claimed that it was an intended feature to reward “system mastery,” for example here. He credits/blames Wizards of the Coast’s familiarity and success with system-mastery-based rewards in Magic: the Gathering for this choice.
But this was written well after the fact, and always struck me as an attempt to save face in hindsight. I’ll buy that there was some intent to make more complex classes have a higher power ceiling, as a way of incentivizing more complex classes and reward those who mastered that complexity, but I don’t for a minute believe that it was intended that the gap be so utterly massive. We know for a fact that, at least early on, Wizards badly misjudged the relative value of certain class features, as admitted by Mike Mearls when discussing ways to improve the very-weak hexblade class.
To the contrary, 4e was written the way it was in reaction to the imbalance of 3e. D&D had never been particularly concerned with “balance,” but with 3e, two things happened:
A new audience more concerned with balance than previous audiences was paying a lot more attention to failures in that realm.
The balance of 3e was really really bad.
So 4e was written to react against that, to provide a more stable, predictable environment to run games of D&D. There were philosophical, game-design reasons for this (belief that balance intrinsically made for a better game), and there were also important business reasons for this (before 4e, book sales were heavily affected by what DMs would allow, which was heavily affected by balance—4e tried to make allowing everything the norm, to improve sales, which they could only do if everything was safe to use without DM vetting).
And then Pathfinder exists primarily as a reaction against 4e, in part against 4e’s devotion to balance. I do not believe that Paizo understands or acknowledges the vast power variance in their own system, but more importantly, they actively disdain the concept of balance in the first place. 4e changed a whole lot of things that some felt made D&D what it is, and many felt it wasn’t “real D&D”—and that includes Paizo and its fans in large part. So I don’t think that the imbalance of Pathfinder was intentional at all—I don’t think Paizo would even agree it exists, though they are objectively wrong on that point—but I do think that not caring about, or even actively disdaining, balance was intentional in Pathfinder.
While we’re on the subject, 5e seems to have been written in reaction to 4e and Pathfinder—namely, in reaction to the fracturing of the fanbase caused by 4e, that Pathfinder (and to a lesser extent, OSR games) took advantage of. 5e was an attempt to recapture the fans who disliked 4e and moved to Pathfinder (or OSR). It doesn’t have the active disdain for balance that Pathfinder does, but it definitely does not hold balance as the chief standard by which an RPG should be judged, as 4e arguably did.
And finally, a note on sources: I am a freelance third-party developer for Pathfinder. I have had to follow Paizo’s development and commentaries closely, and I have had to judge their results, in order to function in that role. That is my basis for the claims I make about Pathfinder’s balance and Paizo’s opinion of the very concept.