Update: different approach to skipping persistent BenchmarkTimers Old way didn't work because of pauses between benchmark runs (global systems would step and recreate them). Tests: ran in editor, confirmed output didn't contain specific timers