[llvm] [AArch64][Codegen]Transform saturating smull to sqdmulh (PR #143671)

[Nashe Mncube via llvm-commits]

nasherm wrote:

Ping

I have made changes to AArch64ISelLowering to enable this transform for `v4i32` types if someone could take a look? I believe this can be extended to `v8i16` types but would require a bit more work as clang generates a different IR pattern when the type is `v8i16`. 

As for SVE I haven't had bandwidth to look at enabling this so I don't currently know if this would be possible/beneficial for an SVE equivalent pattern

https://github.com/llvm/llvm-project/pull/143671