These are needed for when the VIZ experiment is turned on, to evaluate whether tab capture performance remains adequate.

Current thinking is to have a UMA histogram where FPS is measured while mirroring animated content (e.g., videos). In the wide aggregate, we'd expect the "major bars" in the histogram to be the same.

Other UMAs may make sense. We would need to think about other things that would make good "aggregate among all users" measurements for meaningful comparison.