[llvm] [AArch64] add optimisation for MATCH/NMATCH instruction (PR #108179)

[via llvm-commits]

https://github.com/SpencerAbson commented:

This LGTM, I think we can also include [`svhistcnt`](https://developer.arm.com/architectures/instruction-sets/intrinsics/#f:@navigationhierarchiessimdisa=[sve2,sve]&q=histcnt) in this work (here or in another patch)?

https://github.com/llvm/llvm-project/pull/108179