SWDEV-508279 - Improve HIP event profiling

There are 2 functional changes to this patch:
* Use GPU timing for internal markers for HIP.
* Measure CPU time closer to GPU timer, to reduce delta between GPU/CPU timestamp measurements.

There are some smaller non-functional updates:
* waifForFence -> waitForFence typo
* Remove unused drmProfiling

Change-Id: I4c5fa600a842ab60e454888779edcac8449a902a


[ROCm/clr commit: 179801a750]

projects/clr/rocclr/platform/commandqueue.cpp

@@ -251,7 +251,7 @@ void HostQueue::loop(device::VirtualDevice* virtualDevice) {
     // Submit to the device queue.
     command->submit(*virtualDevice);
 
-    // if this is a user invisible marker command, then flush
+    // if this is a user invisible marker with a waiting event, then flush
     if (0 == command->type()) {
       virtualDevice->flush(head);
       tail = head = NULL;