[Profiling] ETW Tracing support ONNX Runtime MLAS/DML and EPs

Is your feature request related to a problem? Please describe.
ORT has several profiling capabilities - e.g. json dump but it missing capabilities where result can be used in xperf. WinML API has logging capabilities which helps IHVs to analysis applications without source code or instrumentation capabilities. We would like to see similar capabilities for application using onnxruntime API. Adding to GUID helps to understand end-to-end overhead when application is running and interaction with hardware.

System information

• ONNX Runtime version (you are using): ORT1.11.1

Describe the solution you'd like
Adding GUID as default in ORT for profiling where developer can see time in operators, binding and session create. WinML profiling solution helps to view result in ETL (generic events) - https://github.com/microsoft/Windows-Machine-Learning/blob/master/Tools/WinMLRunner/README.md.
logman start winml -ets -o winmllog.etl -nb 128 640 -bs 128
logman update trace winml -p {BCAD6AEE-C08D-4F66-828C-4C43461A033D} 0xffffffffffffffff 0xff -ets
WinMLRunner.exe -model C:\Repos\Windows-Machine-Learning\SharedContent\models\SqueezeNet.onnx -DebugEvaluate
logman stop winml -ets

Describe alternatives you've considered
Json profiling but need model access to start with performance analysis.

Additional context
Add any other context or screenshots about the feature request here.

Start by reviewing ONNX Runtime's existing JSON profiling capabilities and the linked WinMLRunner ETW workflow. Define how tracing should expose operator, binding, and session-creation timing in ETL output, then verify the result with an equivalent logman capture and an ONNX Runtime model run.