0024 authorship-before-structural-quarantine (L1) + 0025 relation-vocabulary enforcement (L2) — COUPLED round-__ROUND__ external-review package (__VERSION__)
=====================================================================================
package built:   __TS__ (UTC)
source commit:   __COMMIT__   (PACKAGE_MANIFEST.txt names the same one; the
                 sealer REFUSES if they differ)
__CANDIDATES__
                 __HISTORY__
                 (C5-1: history and the what-changed pointer are DERIVED
                 from the review/package ledger and re-derived at
                 verification — this line first seals under the accepted
                 C-plus construction with this template.)

__CHANGES__

                 Dependency declarations, DERIVED at seal time from each
                 spec's own header line (round 2 package correction — the
                 previous hand-typed sentence drifted the day 0024's
                 header changed):
                 __REQUIRES__
                 Coupled in ONE archive for review economy — per-spec
                 verdicts requested; 0024's ACCEPTANCE waits on 0025's
                 (its Spec-Requires names it) while the implementation
                 freezes stay SEPARATE: L1's is a measurement constraint
                 (the paper-2 probe suite scores the conflation it fixes).

WHAT EACH SPEC IS, in one sentence each:
  0024 (L1)  `_disclosure_for` tests the RELATION before the AUTHOR, so a
             user-authored event whose triple the extractor mislabels
             `third_party_claim` is quarantined as though a third party said
             it (the measured figures live in the spec's §1 and recompute
             from specs/evidence/0025/corpus_aggregate.json — PACKAGE-R6-1
             removed the duplicated number this header carried, the same
             second-copy class as the round-1 counts). The fix removes the model's
             power to unilaterally demote user testimony; the genuinely
             third-party relay stays quarantined. The invariant list is
             §6's table IN THE SPEC — not restated here, where it drifted
             once already (round 1 found three versions of this surface,
             and this line's hand-typed count was one of them).
  0025 (L2)  34.9% of extracted triples are off-vocabulary; they resolve to
             None at `graph.py`'s registry lookup, so they can NEVER
             supersede and never enter contention — restatements accumulate
             as parallel edges. Enforcement is polarity-first with the
             reserved `unclassified` member pinned as a module constant.
             The invariant list is §6's table in the spec — not
             restated here (same round-1 lesson).

MEASUREMENT CONTEXT — GENERATED, not promised:
                 __CONTEXT__

REPRODUCTION — a launcher, and it REFUSES rather than reporting:
  bash specs/evidence/offline/run_offline.sh

  It uses $VERACIUM_PYTHON (or python3), creates a fresh venv (an existing
  one is REFUSED with the creating interpreter named), installs the pinned
  wheels with --no-index --require-hashes, then asks THIS REPOSITORY whether
  the runtime is qualified — `runtime_supported()`. If that is not exactly
  True it EXITS 2 without running the suite.

  LAUNCHER RESULT ON THIS PACKAGE'S FINAL STATE: __LAUNCHER__

measured line:   __MEASURED__ (exit 0)
                 TWO-PHASE, described honestly: the suite is measured FIRST,
                 in the committed tree at the commit above, and THIS FILE IS
                 BUILT FROM THAT RUN afterwards.
                 What IS verified with the file present, RUN FROM THE
                 EXTRACTED ARCHIVE before staging — this list is GENERATED
                 from the sealer's registry and the sealer aborts if it runs
                 anything else:
                 __EXTRACTED__
                 The full -rs output ships as COLLECTED_pytest_rs.txt and the
                 decomposition below is COMPUTED FROM IT.

harnesses:       __HARNESSES__
                 (RUN by the sealer on the tree being sealed, and re-run from
                 the EXTRACTED archive before staging. They are the 0022
                 acceptance harnesses — the tree's standing evidence, kept
                 green; neither is an L1/L2 artifact.)

evidence:        __EVIDENCE__

ENVIRONMENT-CONDITIONAL SKIP INVENTORY — GENERATED from specs/skip_inventory.py:
