[llvm] [DAG][X86] Avoid slow masked-store folds on Zen and x86-64-v3 (PR #214187)

[via llvm-commits]

222rohan wrote:

> > @RKSimon can we add this flag to `-x86-64-v3` as well? This will significantly help Zen subtargets which regress as mentioned in issue #213195
> > cc @phoebewang
> 
> Sounds good to me. We tend to tune for the worest one.

Done! added

> please can you update the summary to reflect the latest patch - you're not adding a new hook anymore, just using isTypeDesirableForOp

Done!

https://github.com/llvm/llvm-project/pull/214187