[llvm] [AArch64] Add lowering for `@llvm.experimental.vector.compress` (PR #101015)

[David Green via llvm-commits]

davemgreen wrote:

For i8/i16 would it be possibly to expand the vectors to multiple i32 vectors, perform the compact, then shrink the result down again? It might not be the prettiest/fastest code, but should hopefully allow them to be supported.

https://github.com/llvm/llvm-project/pull/101015