[llvm] [X86] Don't shrink VEX3 to VEX2 on a symbolic compare predicate (PR #213172)

[Zane Hambly via llvm-commits]

Zaneham wrote:

Thanks again @phoebewang, if you have some time would you be able to merge this for me? 

https://github.com/llvm/llvm-project/pull/213172