It looks like there's no MSE usage in that file.

MSE demux (+clockless rendering) therefore could be missing some relatively simple-to-add performance telemetry.

Notes to self about related, but distinct perf test coverage for MSE:
* crouleau@ has "BusyJS" which can help with assessing other things like sourceopen latency while page is "keeping busy"
* tough_video_cases has "time_to_play" which is even larger-granularity perf measurement.
* see also mse_cases, which might be redundant with pi_perftest if pi_perftest contained MSE cases, (but with more infra overhead -- does it also have more coverage than pi_perftest could provide??)

Marking P2, M-63 for now, but may get bumped due to other priority work.