[llvm] [LICM] Do not reassociate constant offset GEP (PR #151492)

[Yingwei Zheng via llvm-commits]

https://github.com/dtcxzyw approved this pull request.

Make sense to me. However, if the gep chain gets reassociated and idx2 is folded into a constant later, we may still miss some CSE opportunities. See also https://github.com/dtcxzyw/llvm-opt-benchmark/pull/2626#discussion_r2245874120.

https://github.com/llvm/llvm-project/pull/151492