[llvm] AMDGPU: Fix assert when multi operands to update after folding imm (PR #148205)

[Juan Manuel Martinez Caamaño via llvm-commits]

jmmartinez wrote:

It seems it's the same issue as https://github.com/llvm/llvm-project/pull/148187 .

You can avoid the extra `SmallVector` if you iterate in the same way as the PR above (or at least use a Set (I think SetVector to preserve the insertion order when iteration) to avoid scan through is_contained.

https://github.com/llvm/llvm-project/pull/148205