[llvm] [AMDGPU] Limit unclaused VMEM w/a to specific targets (PR #209843)

[Krzysztof Drewniak via llvm-commits]

https://github.com/krzysz00 commented:

... is this correct? If a feature is present in gfx125X for all X, it should be in the generic one?

https://github.com/llvm/llvm-project/pull/209843