[llvm] [AMDGPU] Factor out common code from SIInsertWaitcnts (PR #83018)

[Christudasan Devadasan via llvm-commits]

cdevadas wrote:

Since this patch does code refactoring and modifies SIInsertWaitcnts.cpp considerably. Shall I also use the AMDGPU prefix to the filename instead of SI?

https://github.com/llvm/llvm-project/pull/83018