Controlling Stats Window Placement via API
Automated testing of the dynamic polling will involve comparison with a time sequence "gold" file. Because it is impossible to reproduce the exact same poll timing every run, there will be a small variation in the values. This may make it difficult to determine whether any given test run truly "passes" (some leeway will need to be built in, but this creates a tradeoff between test brittleness and false negatives).
If we can programmatically control the offset of the sliding window every poll iteration, we could likely generate highly reproducible results.
Attempted adding this functionality to the middleware, but trial run is still not 100% consistent. Noticed some unexplained adjustments being made to the polling time based on the newest frame in the ring and also based on some values being stored in a map and mutably referenced. Better to proceed with this with assistance of the original author. Might also require injected stepping of PresentData ETW processing (though in theory this should not be necessary).