Negative search questions need retrieval-only semantics until an answer stage exists

Reusable benchmark constraint from e3-solutions/sesh / COR-4025 (observed 2026-09-04, commit 932a5e3): a top-k retriever cannot honestly report answer abstention or answer false-positive rates. For synthetic no-answer controls, keep them generated and pilot-only, adjudicate the returned top-k, and report only whether those retrieved candidates contain relevant/supporting evidence. A positive hit is a contradiction that invalidates the negative premise; no hit proves only the adjudicated top-k, not corpus-wide absence. Reportable negative evaluation requires independent corpus-wide verification plus a real answer/abstention stage.