The #1495 regression test required 2,000 receiver calls to finish within
2 s. CI shards run with v8 coverage on slower runners: the test took 1.9 s
on #1517's run and 3.2-3.6 s on main afterwards, failing every PR.
It now compares the fastest of two runs at 1,600 and 400 calls. That
ratio is about 5-6x for the fix, with or without coverage, and about
13-14x for the quadratic code before #1517, so the bound of 10 holds on
any runner while still catching the regression.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
