[llvm] [AArch64][Machine-Combiner] Split loads into lanes of neon vectors into multiple vectors when possible (PR #142941)

[Amara Emerson via llvm-commits]

aemerson wrote:

@jcohen-apple Given that we have seemingly 2 separate issues from committing this (the machine verifier errors and the miscompile in https://github.com/llvm/llvm-project/issues/150004) we should consider reverting this to get the tree back into a working state. If it was just the verifier errors I'd expect a quick response on #149585 to be enough, but it doesn't look like we have a fix yet for the miscompile.

https://github.com/llvm/llvm-project/pull/142941