[clang] [Clang] speed up -Wassign-enum via enumerator caching (PR #176560)

[via cfe-commits]

https://github.com/Sirraide approved this pull request.

Changes lgtm.

(I was deliberating whether we should add a stress tests w/ e.g. 10000 enumerators that demonstrates that this is actually faster, but I’m not sure we have the infrastructure in place to actually test for that—at least not in CI afaik—so I think this is fine; doing a binary search and caching enumerator values seems like a better way of doing this regardless.)

https://github.com/llvm/llvm-project/pull/176560