README: status section updated - CI complete, evaluator contracts restored, 30/34

Author: RedEyeNinja-BKK

README.md

@@ -126,9 +126,8 @@ A formal v1.9.0 release qualification bundle (`release-v1.9.0-r1`) has not yet b
 Key items remaining before v1.9.0 qualification:
 - **Scenario evaluator** — the current regex-based runner (`scripts/run_scenarios.py`) is a grading prototype, not an execution harness. It must be split into capture / evaluate / finalize stages with semantic judging or explicit human review.
 - **End-to-end package trials** — at least three complete packages (lightweight noncoding skill, standard persona+skill package, high-assurance production package) built from intent through deployment.
-- **CI expansion** — provenance adversarial tests, evaluator mutation tests, and scenario schema validation must run in CI.
 
-The v1.8.2 qualification evidence is preserved at `evals/runs/release-v1.8.2-r1/`. The bundle records 34/34 under the current first-response evaluator contract. Transcript identity and provenance integrity are mechanically verified; evaluator-contract revalidation remains pending — several contracts were relaxed during terra alignment and require negative-fixture restoration before the contract itself can be considered fully validated.
+The v1.8.2 qualification evidence is preserved at `evals/runs/release-v1.8.2-r1/`. The bundle records 30/34 under the restored evaluator contracts. Transcript identity and provenance integrity are mechanically verified. The 4 honest gaps per run (2 semantic judge, 2 contract) are documented in the bundle manifests.
 
 ## Using Process Engine