[llvm] [AMDGPU] Remove redundant S_AND handling from optimizeVccBranch (PR #228417)

[Jay Foad via llvm-commits]

jayfoad wrote:

> Could it make sense to keep some test cases (with reversed expectations for the positive tests) to document that those patterns are no longer handled by the pass?

I thought about that and decided not to do it. We don't generally have tests for all the optimizations that are not done by a particular pass.

https://github.com/llvm/llvm-project/pull/228417