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

[Alex MacLean via llvm-commits]

https://github.com/AlexMaclean commented:

Nice, this seems like an overall improvement, barring some minor nits.

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