[llvm] [NVPTX] Enhance `mul.wide` and `mad.wide` peepholes (PR #150477)

[Justin Fargnoli via llvm-commits]

justinfargnoli wrote:

> My main concern here is that this seems be be incorrect. If any wrapping occurs in the multiplication won't that no longer occur with the mad.wide?

Thanks for pointing this out. I believe that [these checks](https://github.com/llvm/llvm-project/blob/a69e982ac014a5a957c722f1fea62eb26c7dedeb/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp#L5428-L5434) should address this issue.

https://github.com/llvm/llvm-project/pull/150477