The issues here have rarely ever been that there are compositions that better synergize or otherwise outperform others, so long as that margin is small. It's that those who have no actual expertise by which to assess those merits see a way to simplify the game knowledge or adaptation required of them, and use the meta as an excuse to push a smaller pool of gameplay possibilities, making the overall game easier for themselves.

This amounts to "I don't want to deal with having to deal with anything that's not already in my muscle memory or not among the skills I'm used to tracking, so **** off." That the composition is some 3% inferior, less than a standard deviation driven by crits or mechanical RNG, is merely excuse in the majority of cases. The only real exception to this outside of the game's most challenging content are when there is objectively less community experience with a given unconventional composition, such that it is very unlikely that the party will be able to push those toolkits for all their worth, and the leader isn't willing to spend that kind of practice time — and that's fully his right to want the team to push through without any hitch outside those presented by the fight itself. Though the composition in most cases will cause only a very small portion of the time necessary to clear unless the fight makes certain skills not found in said composition particularly useful.

You'll see the same issue in Overwatch, where people treat compositions as if they were continuously effective typal cards — this somehow being seen as countering that, even if pushed 30 meters away from being able to interact with the character he's supposed to be countering, by yet another character he's supposed to hard counter — instead of actual players with cooldowns or resource limitations on each skill in their toolkit. It's not as if even a Doomfist, a character classically "weak" against Bastion, can't allow for a team to push in against a Bastion, if but briefly enabled — knocking the Bastion out of Sentry Configuration regardless of any barriers placed in his path, and one-shotting the Bastion against Orisa if she's foolish enough to try to save him but a half-second late. People are simply unwilling to figure out how they can support or be supported by a Doomfist, because it takes a slightly new frame of reference by which to identify windows of opportunity, and they don't want to deal with it. To an extent, that's sensible; why compete on a character that is poorly understood as of yet. But at some point it begins to have far more to do with laziness than it does optimization.