Trying to touch some of your points in the most logical order I'll start with your fighter popularity. in 3.x the base classes were just that & pretty much everyone jumped off to some PrC by 5 or so, trying to compare the popularity of any one base class to the 75654754344654 PrCs & claim it as a meaningful data point is a bit odd. In the context of my being asked if "any edition of d&d avoided these issues" though there were plenty of 1/1 & 2/3 BaB classes & PrCs that got multiple attacks during the span of a normal campaign.
That common occurrence of +6/+1 BaB & better on PCs makes how the iterative attack penalty played out & interacts with system math relevant. The way that played out is very different from your initial asserrtion
of "it's not fun for players to keep missing" because it served a very different function of making later attacks that roll well an exciting rather than mundane thing. Now 5e swings the other way with the odds stacked & narrowed so far in favor of success that even getting lucky on on rolls tends to be squarely in the same realm of that twilight zone episode I referenced earlier.
That stacking of the odds & narrowing of chance has direct negative effects limiting the GM's ability to tune the game for the group they have if that group differs too much from bounded accuracy's expected 2-3 player tier1 to low-mid tier2 group of PCs.
I agree that most groups would heal up using spells but doing that left the group a bit vulnerable & wasn't something the group felt comfortable trying to dare the gm when told that taking a rest "here
" seems like a bad idea. System differences came into play at that point. I do not however think that the problems with 5e's math are a personal opinion though because it's an objective fact that bounded accuracy was designed primarily for a certain groupsize & level range. From there it's easy to show how other design changes make it more difficult for a GM to adjust the core math to slide back into being bounded for the group of players & PCs. If
5e came with a very narrow band of levels & said that a GM should never have more than X players because of BA then we'd be talking about ways for the GM to stay within it rather than the problems that arise & needless problems that arise in shifting the bounds... but it doesn't say that for obvious reasons & the GM is left attempting to correct it in a system hostile to correction