Measure reproducible runtime regressions with advisory comparisons.
Measure and investigate runtime regressions with repeated comparable browser scenarios and source-bound evidence. Use for responsiveness, rendering, load, and size changes; not forecast accuracy or release approval.
