✔ frozen numeric core agrees with all 37 retained isolated scalar fixtures (12.290167ms)
✔ independent subject tables are monotone, bounded and 50% at the stated stationary MT (1.543591ms)
✔ random draws are reproducible, in the open unit interval and empirically centered (20.737682ms)
✔ paired 20-pulse endpoints exactly reproduce a separate 20-pulse policy (489.725741ms)
✔ no pulse exceeds the selected ceiling and early stops stay unresolved (2.400425ms)
✔ drift and deliberately forced initial responses have the defined timing (0.220392ms)
✔ review status does not turn all-one-type or imbalanced histories into accepted results (0.925438ms)
✔ simulation cannot silently enable treatment transfer (0.163827ms)
✔ design counts and random-response mixture retain their intended meaning (1.391321ms)
ℹ tests 9
ℹ suites 0
ℹ pass 9
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 607.105817
