[Parameter Capturing] Make probe management APIs asynchronous

[schmittjoseph]

Summary

This PR converts the probe management APIs to be asynchronous. The prior behavior was to submit requests to our profiler without any feedback on if the requests succeeded or failed. Now the results of requests are reported back to our managed code so we can act on it accordingly.

High-level changes:

• Move the probe faulting behavior from the mutating profiler into ParameterCapturingService.
• Support registering probe management callbacks with the mutating profiler. These callbacks will be invoked by a new dedicated thread (See ProbeInstrumentation::InitBackgroundService for an explanation of why this is necessary).
• FunctionProbesManager now maintains an internal state to properly handle callbacks and transition its new task completion sources as necessary.

Out of scope of this PR:

• The dotnet-monitor side changes for logging probe faults

Release Notes Entry