Add device synchronization before destroying proxy thread. (#1631)

This commit ensures that GPU finishes all kernel before destroying
communicator thread.

[ROCm/rccl commit: 52654e2301]

projects/rccl/src/proxy.cc

@@ -1702,6 +1702,7 @@ void* ncclProxyService(void* _args) {
   }
 
   // Wait for all operations to complete and stop progress thread before freeing any resource
+  hipDeviceSynchronize();
   if (ncclProxyProgressDestroy(proxyState) != ncclSuccess) {
     WARN("[Proxy Service] proxyDestroy failed");
   }