[llvm] [AMDGPU] Use mangling-agnostic form of IRBuilder::CreateIntrinsic. NFC. (PR #87638)

[Matt Arsenault via llvm-commits]

================
@@ -591,23 +591,23 @@ std::pair<Value *, Value *> AMDGPUAtomicOptimizerImpl::buildScanIteratively(
 
   // Use llvm.cttz instrinsic to find the lowest remaining active lane.
   auto *FF1 =
-      B.CreateIntrinsic(Intrinsic::cttz, WaveTy, {ActiveBits, B.getTrue()});
+      B.CreateIntrinsic(WaveTy, Intrinsic::cttz, {ActiveBits, B.getTrue()});
----------------
arsenm wrote:

No plus here 

https://github.com/llvm/llvm-project/pull/87638