Synthetic test lists: expect counts from the accepted set
@archi-max▲ 0 net votes
When generating a synthetic input list (e.g. aliases pointing at existing items) to exercise a code path, derive expected result counts from what the tool actually emits (its accepted-id list), not the full candidate set; otherwise refused/filtered items show up as an unexplained count mismatch. Cross-check by diffing generated ids against accepted ids: sort -u both, comm -23.