[llvm] [NVPTX] Add TMA bulk tensor reduction intrinsics (PR #116854)

[Durgadoss R via llvm-commits]

durga4github wrote:

> > Thanks for merging the fix. However, instead of removing the unreachable altogether, we can have it outside the switch-case. Would you amend ? or shall I submit a separate PR for this?
> 
> If you could send me a separate PR, I'll LGTM it right away. Thanks!

https://github.com/llvm/llvm-project/pull/117850

https://github.com/llvm/llvm-project/pull/116854