fix(OMN-18595): release changelog generator strips bare OMN ticket refs

[jonahgabriel]

Fixes the release train CHANGELOG generator at the source so the next
scheduled cut clears the omnibase_core doc-content scan without a hand-applied
per-line suppression marker.

Problem

render_changelog_entry in scripts/ci/release_train.py embeds each
unreleased commit squash-merge subject verbatim into the release PR
CHANGELOG block. Every subject in this fleet routinely carries a bare
OMN-<digits> reference -- as the conventional-commit scope
(feat(OMN-19339): ...) or parenthetically (... (OMN-18157) (#1745)).
CHANGELOG.md is not under onex_change_control/ or contracts/, so the
omnibase_core doc-content scan TICKET_REFERENCE rule
(omnibase_core.validation.doc_content_scan.handler) flags every such line.

Confirmed live: omnibase_core#1746 (0.47.23, opened by the train
2026-09-24) hit this and was fixed by hand-appending the scan
doc-content-ok suppression marker to the two new CHANGELOG lines on the
release branch. That fix does not survive the next cut -- every future
release PR hits the same red until the generator itself changes.

Fix

_strip_ticket_references drops the bare OMN-<digits> token from each
commit subject before it is written into the CHANGELOG, handling both the
scope form and the parenthetical form. The PR-number citation
((#nnnn)) is left untouched -- this mirrors the precedent every hand-cut
CHANGELOG entry already set: PR-number-only traceability, no ticket id
(the existing v0.47.22 entries in omnibase_core/CHANGELOG.md are the
example).

This does not touch the scan or its escape hatches (doc-content-ok, the
onex_change_control/contracts path exemption) -- the check is correct for
a public-facing doc like CHANGELOG.md, so the generator is what moves.

Also declares the one pre-existing undeclared config/migration_classes.yaml
entry (.../0045_terminal_construction_outcome_metrics.sql, landed by an
unrelated PR, #3945) that was blocking every commit in this repo via the
always_run: true onex-check-migration-class pre-commit hook -- using the
checker own --suggest reading, unrelated to this change but landed in the
same PR per the pre-existing-violations policy.

Proof

Focused local tests, launching host (omnibase_infra, pytest -n0, not
directory-wide):

tests/scripts/test_release_train_changelog_ticket_refs.py .... 6 passed
tests/scripts/test_release_train_nightly.py ........................... 72 passed
tests/scripts/test_release_train_lab_ancestor_premise_omn18664.py ..... 18 passed

RED reproduced against the REAL scanner before writing the fix (not asserted
from the diff): loaded the pre-fix scripts/ci/release_train.py (via
git show origin/dev:...) and ran its render_changelog_entry output
through omnibase_core.validation.doc_content_scan.handler.scan_source --
flagged=True, two TICKET_REFERENCE findings (OMN-19339, OMN-18157).
Same call against the fixed generator output: flagged=False. A third
call, feeding the scanner an unstripped line directly, still flags it
(positive control -- the scan was not weakened).

uv run mypy scripts/ci/release_train.py --strict -- clean.
uv run ruff format/ruff check --fix on both touched files -- clean.
pre-commit run --all-files (via the commit) -- green after the
migration_classes.yaml entry above.

This is a scripts/ci/ tooling change with no runtime/src surface --
lab-first (rule 24) does not apply; nothing here runs in a deployed lane.

Evidence-Ticket: OMN-18595

Evidence-Source: OCC#11058

[onexbot-occ-writer]

No OCC evidence companion was minted for this PR.

this PR is already bound to OCC#11058; its evidence companion exists and nothing needs authoring

To clear this: Nothing to do — the companion already exists.

Reported by node_occ_companion_effect (decline code already_bound, OMN-16665). This decision was made against the PR's LIVE state at compute time, not at publish time — a green publisher job only means the command reached the broker.

[github-actions]

Hostile Reviewer — adversarial findings (OMN-17492)

Models succeeded: glm-review
Models failed: codex
New finding threads: 0
Deduped (already posted on this PR): 0
Below quorum (one model only, reported not threaded): 4
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.

Below quorum: 4 finding(s) raised by one model only (OMN-18479)

These are reported and NOT dropped, but they get no thread and do not block: a single model's finding no other model reproduced is not evidence enough to stop a merge. Read them; act on them if they are right.

• [MAJOR] scripts/ci/release_train.py:_SCOPE_TICKET_REF (glm-review) — Scope regex fails for ticket-first multi-token scopes | _SCOPE_TICKET_REF only collapses scopes where the ticket is the sole token or the final comma-separated token. For 'fix(OMN-1234,ci): ...', the
• [MINOR] scripts/ci/release_train.py:render_changelog_entry (glm-review) — Subject reduced to empty bullet when ticket ref dominates | _strip_ticket_references can produce an empty or nearly empty string (e.g. subject 'OMN-4242' or 're OMN-4242' after bare-strip leaves 're')
• [MINOR] tests/scripts/test_release_train_changelog_ticket_refs.py:_scan_source (glm-review) — Hard import-time dependency on omnibase_core in test module | The test imports omnibase_core.validation.doc_content_scan.handler lazily via _scan_source, but there is no skip marker for environments w
• [MINOR] scripts/ci/release_train.py:_strip_ticket_references (glm-review) — Silent mutation of commit subjects with no opt-out path | The change rewrites release-history subjects mechanically. Any ticket id appearing inside a quoted string, path, or otherwise user-meaningful

[github-actions]

⚠️ Hostile Reviewer — DEGRADED (informational)

Critical findings: 0
Major findings: 2
Total findings: 8
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)

[github-actions]

Hostile Reviewer — adversarial findings (OMN-17492)

Models succeeded: glm-review
Models failed: codex
New finding threads: 0
Deduped (already posted on this PR): 0
Below quorum (one model only, reported not threaded): 7
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.

Below quorum: 7 finding(s) raised by one model only (OMN-18479)

These are reported and NOT dropped, but they get no thread and do not block: a single model's finding no other model reproduced is not evidence enough to stop a merge. Read them; act on them if they are right.

• [MAJOR] scripts/ci/release_train.py, _SCOPE_TICKET_REF (glm-review) — Scope regex fails on ticket-first multi-token scopes, leaving malformed output | The docstring and comment claim "fix(OMN-1234,ci): ..." collapses to "fix: ...", but _SCOPE_TICKET_REF requires the OMN
• [MAJOR] tests/scripts/test_release_train_changelog_ticket_refs.py, _scan_source and TestRenderChangelogEntryClearsTheScan (glm-review) — Test module hard-couples to an external package with no skip or isolation | _scan_source imports omnibase_core.validation.doc_content_scan.handler unconditionally. In any environment without that pack
• [MINOR] scripts/ci/release_train.py, _strip_ticket_references (glm-review) — Bare-reference stripping leaves dangling punctuation and commas | _BARE_TICKET_REF removes only the OMN- token. Subjects like "fix: update OMN-1234, add tests" become "fix: update , add tests"
• [MINOR] scripts/ci/release_train.py, render_changelog_entry (glm-review) — Subject reduced to empty string produces an empty changelog bullet | If a subject consists solely of a ticket reference, e.g. "(OMN-1234)", stripping yields an empty string and the changelog renders "
• [MINOR] tests/scripts/test_release_train_changelog_ticket_refs.py (glm-review) — Module-level import-by-path executes the entire release_train script as a side effect | rt = _load() runs at collection time and exec_module executes the full scripts/ci/release_train.py top level. An
• [MINOR] scripts/ci/release_train.py, _strip_ticket_references (glm-review) — Traceability loss: generated changelog drops internal ticket linkage silently | The change removes the only machine-readable link between a changelog entry and its tracking ticket. PR numbers survive
• [MINOR] tests/scripts/test_release_train_changelog_ticket_refs.py, TestStripTicketReferences (glm-review) — No coverage for multiple ticket references or mixed scope-plus-inline forms | Tests cover one ticket per subject in one position each. Uncovered realistic shapes include a scope ticket plus an inline