[llvm] AMDGPU: Fold bitcasts into readfirstlane, readlane, and permlane64 (PR #128494)

[Matt Arsenault via llvm-commits]

arsenm wrote:

### Merge activity

* **Feb 27, 8:58 AM EST**: A user started a stack merge that includes this pull request via [Graphite](https://app.graphite.dev/github/pr/llvm/llvm-project/128494).


https://github.com/llvm/llvm-project/pull/128494