[llvm] [NewGVN] Fix lifetime coercion (PR #141477)

[via llvm-commits]

ManuelJBrito wrote:

> I think it would be a lot cleaner if we first handled the lifetime intrinsic case separately (because it does not return a pointer) and then the alloc/allocation cases (which do return a pointer). Mixing them makes it very non-obvious whether the interaction between all the conditions is correct.

Thanks, that makes sense. I have updated it.

https://github.com/llvm/llvm-project/pull/141477