[llvm] [SystemZ] Don't lower float/double ATOMIC_[LOAD|STORE] to [LOAD|STORE] (PR #75879)

[Jonas Paulsson via llvm-commits]

JonPsson1 wrote:

Patch updated.

@uweigand Now that the Clang FE does not insert the fp<->int bitcasts, the handling for the resulting bitcasts have been removed from the SystemZ backend (there doesn't seem to be any value at all to having them around on SPEC). @arsenm approved the common code changes earlier, so now it's the SystemZ changes that also need review.

@kaz7 This patch shouldn't affect the VE backend at all, but since SDTAtomicLoad has been changed to also allow FP loads, I had to insert the equivalent MVT specifier in the VE patterns for atomic loads. I used 'iAny' for now, but this could easily be changed. 

https://github.com/llvm/llvm-project/pull/75879