[llvm] [AArch64] Prefer `SADDLV` over `ADDV` for vector mask reductions (PR #183522)

[Paul Walker via llvm-commits]

paulwalker-arm wrote:

I've no objection to splitting the PR and waiting to see if the combine proves itself.  As long as the `SelectArithExtendedRegister` change is sufficient to resolve the issue raise in https://github.com/llvm/llvm-project/pull/183025 then I'm happy.

https://github.com/llvm/llvm-project/pull/183522