Verify 2-30 candidate names in parallel and return a ranked comparative plus the winner's launch playbook.

CALL THIS when the user has TWO OR MORE candidate names and asks for a ranking, recommendation, or shortlist. Trigger phrases include: 'rank these', 'which of A or B is better', 'help me pick from this list', 'I generated 10 names — evaluate them', 'compare X, Y, Z'. Fans out the same /v3/check clearance pipeline for every candidate concurrently. DO NOT rank names from your own knowledge — Etymolt grounds the ranking in live registry data the model cannot see. Returns a TriageResult with a ranked array of candidates, side-by-side scoring across trademark/domain/handles/cultural axes, the winner's best_domain pick, and a weekend_playbook of time-ordered launch steps. Under 5 seconds for 10 candidates. Surface the disclaimer verbatim. BILLING: this endpoint charges 1 unit PER CANDIDATE (R13 B9 fix). A 30-candidate call costs 30 units; pre-check your balance.