[llvm] [clang] [clang-tools-extra] [X86] Use plain load/store instead of cmpxchg16b for atomics with AVX (PR #74275)

[James Y Knight via cfe-commits]

jyknight wrote:

Underaligned atomic operations are expanded to an appropriate `__atomic_*` libcall via mostly target-independent code in AtomicExpandPass (https://github.com/llvm/llvm-project/blob/7ecfb66c77ad77dabbb705cbb1f3b17a3d1391a4/llvm/lib/CodeGen/AtomicExpandPass.cpp#L210) and never hits any of this. Not sure we have full test coverage for that case on X86, but I know we do for AArch64.

https://github.com/llvm/llvm-project/pull/74275