[llvm] Use (de)interleaving intrinsics for fixed-length vectors (PR #202583)

[Luke Lau via llvm-commits]

https://github.com/lukel97 commented:

The AArch64 neon improvements look good, I don't see any regressions.

Would you mind enabling this by default for all targets just so we can see what the impact is on e.g. RISC-V + X86? 

https://github.com/llvm/llvm-project/pull/202583