[clang] [llvm] [clang][OpenMP] Add no-loop SPMD kernel promotion (PR #224041)

[Robert Imschweiler via cfe-commits]

ro-i wrote:

performance analysis is probably mostly a matter of grid size selection, and is thus already covered by me :) See https://github.com/llvm/llvm-project/pull/234660, for example, where I'd like to upstream a no-loop grid size optimization.

https://github.com/llvm/llvm-project/pull/224041