[llvm] [X86] Always use 128-bit V_SET0/AVX512_128_SET0 patterns, along with SUBREG_TO_REG for extension to 256/512-bit vectors (PR #212950)

[Phoebe Wang via llvm-commits]

phoebewang wrote:

I asked AI to write a validation. It proved the two sequence are identical: https://godbolt.org/z/Tj9sxMa89

https://github.com/llvm/llvm-project/pull/212950