[Libomptarget] Add a wavefront sync builtin for the AMDGPU implementation (#70228)

Summary:
While this is technically a no-op for AMDGPU hardware, in cases where
the user would see fit to add an explicit wavefront sync on Nvidia
hardware, we should also inform the LLVM optimizer that this control
flow is convergent so we do not reorder blocks.

Author: jhuber6

openmp/libomptarget/DeviceRTL/src/Synchronization.cpp

@@ -272,7 +272,9 @@ void fenceSystem(atomic::OrderingTy Ordering) {
 }
 
 void syncWarp(__kmpc_impl_lanemask_t) {
-  // AMDGCN doesn't need to sync threads in a warp
+  // This is a no-op on current AMDGPU hardware but it is used by the optimizer
+  // to enforce convergent behaviour between control flow graphs.
+  __builtin_amdgcn_wave_barrier();
 }
 
 void syncThreads(atomic::OrderingTy Ordering) {