pub fn corpus_is_sufficient(evidence: &str, min_corpus_chars: usize) -> boolExpand description
Whether there is enough evidence to judge a candidate against at all (G-PR-7).
Single source of truth for “is this corpus worth grounding against”,
shared by the entity-description write path and the --status quality
sampler. Before this existed, both asked
PreservationVerdict::evaluate_grounding instead, which answers
Preserved { score: 1.0 } when the evidence is empty — so the writer
persisted filler for unbound entities and the sampler counted those same
entities as PERFECT quality. The measurement shared the defect of the
thing it measured, which is why the problem stayed invisible.