[llvm] [ValueTracking] Infer NonEqual from dominating conditions/assumptions (PR #117442)

[Nikita Popov via llvm-commits]

nikic wrote:

Could you please rebase now that the SimplifyQuery change landed? I'm generally inclined to just go ahead with this change, the compile-time impact is minor and this adds useful functionality. Only thing I'd check is if not using isImpliedCondition would help mitigate the impact (and only handle the trivial icmp eq/ne with matching operands case).

https://github.com/llvm/llvm-project/pull/117442