[llvm] [AMDGPU] Disable atomic optimization of fadd/fsub with result (PR #96479)

[via llvm-commits]

b-sumner wrote:

We have very important supercomputer customers waiting for this who are going to be dissatisfied if it only works when the result is not used.  We need an optimized implementation for the returned result case.  If the CTS broken makes this happen faster then I think it should remain broken.

https://github.com/llvm/llvm-project/pull/96479