[llvm] [NVPTX] Generalize and extend upsizing when lowering 8/16-bit-element vector loads/stores (PR #119622)

[Drew Kersnar via llvm-commits]

dakersnar wrote:

Just rebased and reran tests, everything looks good. Should be ready to merge. @Artem-B Can you merge this for me? Thank you :)

https://github.com/llvm/llvm-project/pull/119622