[llvm] [docs] Reword the alignment implications for atomic instructions. (PR #75871)

[James Y Knight via llvm-commits]

jyknight wrote:

Overall, yes.

I don't think the mention of AtomicExpandPass belongs here. It should instead just say something like "Note: if the alignment is not greater than the size, the atomic operation is likely to require a lock and have poor performance."

https://github.com/llvm/llvm-project/pull/75871