[llvm] [InstSimplify] Optimize maximumnum and minimumnum (PR #139581)

[Lewis Crawford via llvm-commits]

LewisCrawford wrote:

Adding @Artem-B as a reviewer, as NVPTXTargetTransformInfo.cpp currently transforms nvvm_fmax.* into maxnum, but should really use the new maximumnum intrinsic instead to match the SNaN semantics of PTX max. However, maximumnum is currently not optimized as much as maxnum, so this patch is trying to add some of those missing optimizations in InstSimplify before NVPTX switches to using maximumnum in a future patch.

https://github.com/llvm/llvm-project/pull/139581