[llvm] [AMDGPU] Fix GCNRewritePartialRegUses pass: vector regclass is selected instead of scalar. (PR #69957)

[Carl Ritson via llvm-commits]

https://github.com/perlfu commented:

I think the code change looks fine.
I left a few questions inline about existing code -- you don't necessarily need to address these.

Could you regenerate the tests as a separate NFC commit (or PR), so this PR only shows test changes from the code change?

https://github.com/llvm/llvm-project/pull/69957