[llvm] [AMDGPU] make AMDGPUUniformIntrinsicCombine a function pass (PR #165265)

[Sameer Sahasrabuddhe via llvm-commits]

================
@@ -97,14 +91,12 @@ static bool optimizeUniformIntrinsic(IntrinsicInst &II,
           Tracker[NotOp] = true; // NOT preserves uniformity
           LLVM_DEBUG(dbgs() << "Replacing ICMP_EQ: " << *NotOp << '\n');
           ICmp->replaceAllUsesWith(NotOp);
-          ICmp->eraseFromParent();
----------------
ssahasra wrote:

I vaguely remember that since you use early_inc_range in the outer for-loop, it should be safe to call this erase, right?

https://github.com/llvm/llvm-project/pull/165265