[llvm] [AMDGPU] - Add constant folding for s_bitreplicate (PR #72366)

[Jessica Del via llvm-commits]

https://github.com/OutOfCache commented:

I implemented the `s_bitreplicate` constant folding, as mentioned in https://github.com/llvm/llvm-project/pull/69209#discussion_r1374324260.

However, I don't think I can do the same for the `s_wqm` and `s_quadmask` intrinsics, because they implicitly set SCC?

https://github.com/llvm/llvm-project/pull/72366