[llvm] [AMDGPU] Add wait states between different MFMAs sharing an accumulator (PR #218363)

[Pankaj Dwivedi via llvm-commits]

PankajDwivedi-25 wrote:

if we don't have any concerns on this can I move forward, as this is blocker for triton.

https://github.com/llvm/llvm-project/pull/218363