[llvm] [X86][Atomic] Fix bts/btr/btc regression caused by InstCombine trunc to i1 (PR #200667)

[Simon Pilgrim via llvm-commits]

https://github.com/RKSimon commented:

If the middle end won't create it should we drop the AND(X,1) handling from the backend and just handle the TRUNC case? Would be nice to remove X86ISelLowering code for cases that we shouldn't see anymore.

https://github.com/llvm/llvm-project/pull/200667