chore(OMN-18202): bump omnibase-core pin to 0.47.12 and carry the three version-bookkeeping surfaces

[jonahgabriel]

Summary

The published omnibase-infra 0.38.22 pins omnibase-core==0.47.9. omnimarket
0.4.62 requires omnibase-core>=0.47.11, so omnimarket's release run
34667694260
failed its PyPI pin-resolvability step (OMN-18034) with the two constraints
mutually unsatisfiable from the real index:

Because omnibase-infra>=0.38.22 depends on omnibase-core==0.47.9 and
omnimarket==0.4.62 depends on omnibase-core>=0.47.11, we can conclude
that omnibase-infra>=0.38.22 and omnimarket==0.4.62 are incompatible.

infra dev already carried 0.47.10 from the v0.47.10 cascade (#3437) but no
release has been cut since, so the published artifact still carries 0.47.9.
omnibase-core 0.47.12 is now on PyPI (uploaded 2026-09-12T01:30:49Z), so this
bumps the pin straight to 0.47.12 rather than to 0.47.11 with a second bump to
follow.

This PR exists because the release dependency-cascade generator does not carry
omnibase_infra's version bookkeeping — OMN-18202 AC2. All three surfaces named
there are updated in the same commit as the lockfile:

Surface	How it was produced
docker/runners/runner-image.lock.json	scripts/ci/runner_image_identity.py --mode generate — deterministic; the digest binds pyproject.toml and uv.lock as version components
src/omnibase_infra/runtime/version_compatibility.py _FALLBACK_MATRIX	confirmed in sync by scripts/update_version_matrix.py --check (exit 0, "Fallback matrix is already up to date")
tests/fixtures/seams/core_release/0.46.11_expected_symbols.json core_version	bumped 0.47.10 -> 0.47.12

Verification

Focused suites over the three affected surfaces, with a negative control proving
each one actually binds to the version rather than passing vacuously:

• All three reverted to 0.47.10: 3 failed, 12 passed, one named failure per
  surface — test_recorded_identity_matches_recomputed_identity,
  test_update_version_matrix_check_passes,
  test_seam_released_core_pin_importable_and_typed.
• All three correct: 15 passed.

uv run pytest tests/ci/test_runner_image_identity.py \
  tests/unit/runtime/test_fallback_matrix_sync.py \
  tests/unit/runtime/test_seam_released_core_pin.py -q

Regenerated digests: identity 9d792733f04ae323f11610aefdffd328 ->
d1eac15983c9c8f03033b9bb3a31bd26, shared env e5b91f9191f06a0d66dcf5cf ->
387313789eda85dd7acdf46b. image_version unchanged at 7.

Governed pre-push ran clean (check-only hooks; the repo's local test leg was
retired in #3415). It reported the deploy-scoped surface
src/omnibase_infra/runtime/version_compatibility.py and that OMN-18202 had no
OCC contract at push time — the companion supplies it.

Follow-on

Once this lands, omnibase_infra v0.38.23 is cut so the published artifact
carries the corrected pin, then omnimarket 0.4.62 is re-run. 0.4.62 never
published: its tag-creation and publish steps were skipped, and PyPI still
serves 0.4.61 as info.version with 0.4.62 absent from releases.

Test plan

• Focused suites over the three bookkeeping surfaces, with negative control
• uv lock --upgrade-package omnibase_core resolved cleanly (0.47.10 -> 0.47.12)
• CI green on the live head

Evidence-Ticket: OMN-18202
Evidence-Source: OCC#9184

[github-actions]

Hostile Reviewer — adversarial findings (OMN-17492)

Models succeeded: glm-review
Models failed: codex
New finding threads: 5
Deduped (already posted on this PR): 0
Nit-level findings suppressed: 1

The model is the FINDER, never the gate: merge is gated only by the
deterministic Hostile Review Thread Gate, which blocks while
hostile-reviewer threads are unresolved. Resolve each thread after
addressing (or rejecting, with a reply) its finding.

Findings demoted from threads (anchor rejected)

• [MAJOR] hostile-reviewer (glm-review)

  Fixture filename and recorded core_version diverge | The fixture file is named 0.46.11_expected_symbols.json but its internal core_version now reads 0.47.12. The update procedure embedded in the fixture's own comments says to bump core, re-run the seam test, and record the new surface 'here', implying the filename should track the version it snapshots. Either the filename is stale from a prior bump (suggesting the seam surface was never actually re-derived against 0.47.12) or the convention is inconsistent.

  Resolve this thread when addressed — the Hostile Review Thread Gate blocks while hostile-reviewer threads are unresolved (OMN-17492).

• [MAJOR] hostile-reviewer (glm-review)

  Symbol surface unchanged across core bump without justification | The diff modifies core_version in the fixture but shows no change to any entry under 'modules'. Core 0.47.10 to 0.47.12 spans at least one release (0.47.11) plus the pre-release merge referenced in pyproject.toml comments (terminal delegation classes from #1439 and runtime_fanout_resolver from #1440). A two-version jump in a fast-moving dependency with zero delta in the expected public surface is implausible unless the fixture only pins a sub

  Resolve this thread when addressed — the Hostile Review Thread Gate blocks while hostile-reviewer threads are unresolved (OMN-17492).

• [MINOR] hostile-reviewer (glm-review)

  Fallback matrix max_version left at 0.48.0 | The fallback constraint raises min_version to 0.47.12 but keeps max_version at 0.48.0. This is likely intentional (semver next-minor ceiling), but the fallback matrix is a duplicate of truth derived from pyproject; the diff does not show any consistency check between the two. If _build_matrix_from_pyproject fails at runtime, the hardcoded fallback silently governs and any drift between it and pyproject produces confusing behavior. | Evidence: _FALLBACK_MATRIX: mi

  Resolve this thread when addressed — the Hostile Review Thread Gate blocks while hostile-reviewer threads are unresolved (OMN-17492).

• [MINOR] hostile-reviewer (glm-review)

  No test changes accompany the dependency bump | The diff bumps omnibase-core by two patch versions including newly merged runtime delegation and fanout resolver code, yet contains no new or updated tests exercising the changed behavior. Passing existing tests does not demonstrate the #1440 runtime_fanout_resolver path is used correctly, only that old tests still pass. | Evidence: pyproject.toml comment: 'pulls the merged core #1439 (terminal delegation classes) AND the merged core #1440 (runtime_fanout_reso

  Resolve this thread when addressed — the Hostile Review Thread Gate blocks while hostile-reviewer threads are unresolved (OMN-17492).

• [MINOR] hostile-reviewer (glm-review)

  Lockfile digests and image_version not cross-verified | runner-image.lock.json updates identity_digest and shared_env_digest but keeps base_image_digest, gh_version, kubectl_version, and image_version unchanged. The digests were presumably recomputed from the new dependency set, but nothing in the diff shows how they were derived, and image_version 7 is unchanged despite the runner image content changing. If image_version gates cache invalidation or deployment pinning downstream, stale consumers may reuse t

  Resolve this thread when addressed — the Hostile Review Thread Gate blocks while hostile-reviewer threads are unresolved (OMN-17492).

[github-actions]

⚠️ Hostile Reviewer — DEGRADED (informational)

Critical findings: 0
Major findings: 2
Total findings: 6
Models succeeded: glm-review

Note: Fewer than 2 reviewer models succeeded. Degraded results are informational (OMN-8468/OMN-8524) and do not block merge. Error: cli_review exit 2 (fewer than 2 models succeeded — partial/total outage)

---

Semantics (OMN-17492 — the model finds, thread resolution gates)

Surface	Meaning	Blocks merge?
Review threads	Per-finding, posted by the reviewer	No (informational)
Hostile Review Thread Gate	Deterministic: unresolved hostile-reviewer threads exist	Fails until resolved (not yet a required context)
degraded verdict	Fewer than 2 models succeeded (infra)	No

Powered by omniintelligence.review_pairing.cli_review — multi-model adversarial review: qwen3-review, qwen3-review-b, glm-review (OMN-8468/OMN-8524/OMN-17492)