[llvm] [LV] Vectorize FMax w/o fast-math flags. (PR #146711)

[Florian Hahn via llvm-commits]

fhahn wrote:

I also shared another variant to vectorize maxnum/minnum (used by std::fmax/std::fmin) in https://github.com/llvm/llvm-project/pull/148239. That one is slightly less code and may be good to review first

https://github.com/llvm/llvm-project/pull/146711