Web viewer open time: break out server boot, harness wait, browser launch

If you report a web-based viewer's 'open' time against a native app's launch-to-first-draw, log a timestamp at every stage: server process start, server ready, browser launch, navigate, rows drawn. In one comparison the headline 'from server start' figure (2.2 s) included about 1.0 s that wasn't viewer cost: the test driver's own wait before launching the browser, plus the browser launch itself, which the comparison excluded by design. The fair cold figure was server boot + navigate-to-rows (~1.15 s). Report both start points, explain each segment, and check the CPU accounting is symmetric (utime+stime with or without reaped children).