[llvm] [clang] [AMDGPU] Adding the amdgpu-num-work-groups function attribute (PR #75647)

[Johannes Doerfert via cfe-commits]

jdoerfert wrote:

We are back to the duplication.
The attribute here, num-work-groups, exist in IR as:
NVPTX: minctasm -> min work groups
NVPTX: maxclusterrank -> max work groups
OpenMP: omp_target_num_teams -> right now exact = num work groups

@bader This is not the same as the intrinsics but we can/should unify the attributes as well, IMHO.

If we make changes in this area, I would strongly suggest a generic attribute and adapted handling in the backends/optimizations.

https://github.com/llvm/llvm-project/pull/75647