[llvm] [AMDGPU] Set Convergent property for image.(getlod/sample*) intrinsics which uses WQM (PR #122908)

[Mariusz Sikora via llvm-commits]

mariusz-sikora-at-amd wrote:

> The code change looks fine to me. However I think for completeness this also needs to be applied to some variants of `image_gather4` as well.

I will prepare new PR for image_gather4

https://github.com/llvm/llvm-project/pull/122908