OClCollMeshEngn: perf profile and print kernel exec durations

stimBuffApis/livoxGen1/openClCollatingAndMeshingEngine.h

@@ -8,6 +8,7 @@
 #include <functional>
 #include <iostream>
 #include <stdexcept>
+#include <chrono>
 #define CL_TARGET_OPENCL_VERSION 120
 #include <CL/cl.h>
 #include <asynchronousLoop.h>
@@ -61,6 +62,10 @@ public:
 		AsynchronousLoop& asyncLoop, StimulusFrame& stimulusFrame,
 		Callback<compactCollateAndMeshFrameReqCbFn> callback);
 
+	// Get kernel execution durations in milliseconds
+	std::chrono::milliseconds getCompactKernelDuration() const;
+	std::chrono::milliseconds getCollateKernelDuration() const;
+
 private:
 	PcloudStimulusBuffer& parent;
 
@@ -103,6 +108,12 @@ private:
 	compactKernelCbFn compactKernelCb;
 	collateKernelCbFn collateKernelCb;
 
+	// Timestamp tracking for kernel execution
+	std::chrono::high_resolution_clock::time_point compactKernelStartTime;
+	std::chrono::high_resolution_clock::time_point compactKernelEndTime;
+	std::chrono::high_resolution_clock::time_point collateKernelStartTime;
+	std::chrono::high_resolution_clock::time_point collateKernelEndTime;
+
 	// Static callbacks for OpenCL events
 	static void CL_CALLBACK compactKernelEventCallback(
 		cl_event event, cl_int event_command_exec_status, void* user_data);