[clang] [llvm] [AMDGPU] Track tensor load/store DMAs with asyncmark (PR #200775)

[Tony Varghese via cfe-commits]

tonykuttai wrote:

The documentation build is currently failing with:
```
/home/runner/work/llvm-project/llvm-project/clang-build/tools/clang/docs/AMDGPUBuiltinReference.rst:402:undefined label: 'amdgpu-async-operations'
```
Seems this PR #200775 added a reference to `:ref:amdgpu-async-operations` in `clang/include/clang/Basic/BuiltinsAMDGPUDocs.td` (line 627), but this label is never defined.

https://github.com/llvm/llvm-project/pull/200775