pm/testing baseline: 52 runs, 4 regressions (v5.17-rc1)