2e's bonuses/penalties for high/low attributes can be the same as, or even in many cases more extreme than, those in d20 D&D.
18 dexterity grants a +4 bonus to AC in both systems. 18 wisdom grants +4 to saving throws in both. A 2e cleric with 18 wisdom gets 6 bonus spells. A d20 cleric gets 4. 18 constitution grants +4 hit points in both systems. In 2e, 18 con allows a 99% chance to survive system shock, 7 con is only 55%, a 44% variation. 18 intelligence grants 7 bonus languages in 2e, only 4 in d20. Wizards in 2e get a lot more for high intelligence than they do in d20, for example an 18 int means an 85% chance to know spells, but it's just 35% for a 9 int, a 50% swing. The bonuses for high charisma are also massive, for example +7 reaction adjustment for an 18 charisma versus +0 at 12.
Oh, I'm not advocating a direct return to 2E or anything. Where d20 was all power treadmills and bland sameness(and I'm not qualified to say what 4E was or wasn't), 2E was often wildly inconsistent, though there was an odd logic to many of the decisions made. I notice you did not mention the limitation on maximum spell level obtainable by intelligence score, for instance. I'm fairly convinced that(again, assuming a 3d6 generation method for adherence to the probability curve) this was intended as an attempt to balance out the power disparity between magic users and mundane classes. Obviously not very effective in practice, but there was an intention behind the decision.
D20, on the other hand, had only a single rationale behind it's math: unity at any cost. Simplify the math. Homogenize it so that you only need to know a single mechanic: that d20 + modifer => DC means success. It wasn't a horrible idea, and it may even been exactly what was needed
at the time. But the idea was flawed. Basing everything on the roll of a d20 was okay(and far preferable to 2nd Edition's additions, subtractions, percentile rolls, etc), but look at what happened. Ability scores are hugely important for class effectiveness, yet paradoxically, upgrading from an even to an odd score means exactly nothing.
Of course, the whole situation was highly exacerbated by the ease with which an ability score could be raised after character generation. Periodic gains through leveling, ubiquitous magic items(that are easily made by a player), etc. In contrast to 2nd Edition, it actually became highly uncommon to see a character that
didn't get at least one stat over 20.
Anyway, mini-rant aside, I do see your point, though a lot of the things you mentioned are not directly related to class effectiveness(and the +4 HP for con only applies to warriors), and some others that are directly related were, IMO, designed to limit casters, especially the chance to learn spells and limited spell level access. All in all, it was a system designed to take advantage of the 3d6 random roll, at least more so that the d20 system ever will be. I don't want to return to it exactly, but I would prefer that the new system take advantage of the curve, rather than sticking to the linear growth that they seem to be eliminating in other portions of the game already.