I’m not saying it applies, here, but there also is a psychological aspect to chess annotations. Annotators may mark a move as good even if they think or even know better moves against optimal play exist. It isn’t fair to expect human chess players to play optimally.
For example, Mikhail Tal’s made multiple sacrifices that won him games even today get a ! annotation from reviewers, even though they aren’t optimal (https://www.chess.com/forum/view/general/how-should-one-go-a...)
I'm no chess player, but that's a cool hobby you have to revisit the game. Throw an analysis up on Medium and post it here, or share it with a chess magazine, and maybe that will be authoritative enough for Wikipedia.
The problem with the Wikipedia article is the subjectivity of how the "?" and "!" annotations on the moves are handed out - sometimes they are quite wrong.
The clearest example in the article is the fact that 37... e6 loses and thus it deserves a double question mark, because 37... e5 draws. Anybody who downloads Stockfish and the six piece table bases can see that very quickly in 2018, but I suppose Wikipedia requires an "authoritative" published source on that before it can go into the article. Similarly on the next move, 38. Rd1 wins, whereas Kasparov's 38. h6 only draws. So that also deserves at least a question mark. (perhaps Chessbase should do an article on it)
Other dubious annotations are 18... f5 when 18... Bd4 is a clear draw (also in the Kasparov and King book) and perhaps 26... f4 when 26... Bc5 was better (in "Reinventing Discovery" the author writes about this move choice quite a bit).