If the point of 5e were parsimony it could have delivered a broader range of possible characters with far fewer classes (as few as 3, I'd say), and the assumption of MCing instead of clumsily designing sub-classes & Backgrounds on the assumption that MCing wouldn't be used.
For that matter,
classless works very well, and is much more efficient in delivering more meaningful choices with less page count.
The point of 5e, though, was to heal the rifts in the community, be inclusive of all past fans (but accessible to new ones), and allow them to evoke the feel of, and play the same sorts of characters they could in, each of those past editions. Among other virtually impossible goals.

That meant making familiar class designs with familiar names, in spite of all the redundancies, inefficiencies and imbalances that entailed.
Given that, there's no slippery slope, no danger of infinite additions to the game, the finite set of past editions' offerings would be the maximum. And, even that'd be a gross exaggeration, as there was a tremendous amount of overlap among those editions. Every class that was ever in a PH1, even every /character type/ that was ever in a PH1, is doable with 2 more classes, the Psion(icist) & Warlord. Every class ever in a PH, would take it up to, what, a half-dozen, perhaps?
A potential, worst-case, 50% increase in player choice (further, gated behind DM opt-in, as with all supplements) is hardly devastating bloat. Really, the current proliferation of sub-classes is more worrisome, that way.
(Though, it is true that a literal PH2 would be against the current conventional wisdom, so it'd have to be another goofy "So-and-so's guide to things you're not interested in, newbie," format.)