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

[Krzysztof Drewniak via cfe-commits]

https://github.com/krzysz00 commented:

Endless bikeshedding aside (sorry about that), I've got a test I'd like to see, but overall, the PR seems reasonable

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