You make a good point and a bad point. The good thing you pointed out is the difference between a game like Super Meat Boy and 8-bit platformers. There is significantly less punishment and therefore less frustration in a game like Meat Boy. The challenge comes purely from the challenge itself, not the fact the game sends you back really far and you have to repeat large sections.
The bad point you made was implying that this is a bad thing.
Allow me to give a rather long-winded, if not an especially educational answer. I think next time I'll remember to quote this, since I've said this over and over.
The "challenge itself" (dumb lol) is the entirety of the "large section" if the game wants you to clear the whole thing without dying.
That is to say the challenge is "do all of this without dying", not "do this" multiplied by w/e amount and any other "limitations" are framed as separate/unnatural/etc. The obvious effect of lowered tolerance for failure (although this may be besides the point) is that it makes the game less vulnerable to trial and error (and sheer luck), i.e. increasing depth(the amount of meaningful complexity (strategic
and reflexive) to master/learn) as seen by a hypothetically measurable "distance" between players(non-mastery vs. high mastery).
Perhaps its necessary to properly define trial and error before I move on: trial and error is the process of repeating attempts, without extrapolating a logic from the results, until you happen upon the correct combination of choices. I say "properly", because "trial and error" is often used to dismiss strictly difficult challenges, when it should only be used for challenges where "skill" (mastery of meaningful complexity) is less of a factor. Scrubs (undisciplined players looking for excuses and escapes) use this because they do not like to reason with difficulty higher than they can handle (impulsively saying: "I can't do it right now, thus it is cheap/punishing/etc."). So a game vulnerable to trial and error demands less mastery than what might first be expected (it could be deceptively easy, large swathes of its complexity lack meaning), with a game that is completely trial and error (limited to it) would be an entirely random game (e.g. requiring coin flips to come up heads five times in a row - highly difficult, but no skill involved). All this can be demonstrated by examining players of different skill levels, with some criteria or another based on the game itself.
However, the larger point here is that you are being entirely arbitrary in what constitutes a "single challenge", which is being used to say a challenge that must be completed before your progress is saved. Moreover, after setting this rule, you treat anything above it as something other than increased difficulty on a spectrum: it becomes "punishment", an essentially negative quality that players should avoid.
Maybe this will help illustrate how arbitrary singling out challenges the way you are is (as well as briefly explain a relationship between strictness and depth): you could easily break down Super Meat Boy, an already heavily "fractured" game, even further than what it is. It would be a simpler matter to take a stage from a "masocore" (dumb lol) platformer as we know it and make each successful jump a stage/"single challenge" in itself (this is already done by scruby players with emulators for older games). You could then even break each jump down into pieces for each general gesture of aerial movement, then maybe even further for every single input and non-input (by the frame)! We could then complain "WTF? The game expects me to repeat inputs after the first time I put them in!?" And as you might expect every single additional input players need to get "correct" (or at least, not "incorrect") in order increases the opportunity for failure (ways to create distance between players) and makes learning solutions more necessary (at least, the time spent using trial and error to find answers, statistically speaking, grows exponentially versus the time it takes to learn the reflexes/strategies required).
Using a term like "punishment" to dismiss certain challenges on a general level (i.e. cowering from a perfectly fine degree of stimulation that hungrier players will rush to enjoy) betrays one's true intention: to shield oneself from "frustration". We must first understand exactly what "frustration" is: the sensation(being reminded) of lacking power ("weakness"). It should go without saying that humans don't like this. However, when faced with something that gives frustration, it may be possible (and, in reasonable matters of skill, it is), if we also have the dedication for it, for us to grow to the point where we are no longer made to feel that way. This results in us growing as players to a visible extent, the obtaining of power in the form of mastery (and this "progression" is much more meaningful and personal than having a "character gain a level"). This satisfaction with oneself is a worthy reason why to take on challenges, if the not the purest reason. More so than the idea of "difficulty", it is "resistance" we seek: force pushing against us so we can sense our own strength as we exert our own force (too much resistance will emasculate us, too little resistance doesn't have much feeling at all). (We can see here how it is resistance that is important, not "achievement" - but that's a tangent.)
To sum it up, metaphorically: There was at first an amount of weight that we could not lift and thus we were painfully reminded of our weakness as we struggled beneath it. Then in our struggle, we grew to the point where we could succeed and in this process we could clearly feel our increased power. If we will not or can not grow, then we are left with that frustration. What is the alternative solution, then? Declare the demand "illegitimate".
See one could even say "the game is too hard (for me)" accepting weakness in the process (though, there is the question how weak they really are compared to other players), but that's a whole different approach than labeling something as "illegitimate". You are not humbled this way, in fact you maybe even feel good about yourself, because it wasn't your fault you failed and you get to feel the satisfaction of laying blame. Going as far to argue "the game is too hard" (or "the game is too luck-based") means having to deal with players who may know better that disagree.
It is very likely that, besides it being good old human psychology, these specific learned excuses for 2D games in particular come from an era of overexposure to the combination of crap flash games and emulator save-state abuse totally fucking up expectations of what how a "hard" 2D game should look like. After all, Super Meat Boy is a perfect blend of these two terrible experiences and its designer(s) embody these ideas.
tl;dr: don't be a scrub
I didn't actually get to reply to the OP like I wanted to(I think there is a good point to be found which will put everything I said in a different perspective), but I guess that will come later because I was exhausted before I started this post... not mention I get the feeling people might put emphasis on the wrong things, but I'm too lazy to explore every possible interpretation right now.