[action-translation] Wire zh-cn sync automation and reconcile drift (Phase 0, Track A)

[mmcky]

Part of Phase 0 tracked in QuantEcon/action-translation#74. This repo is the English source for lecture-intro.zh-cn, which is 49/52 lectures translated and fully human-reviewed, but has had no translation updates since 2026-04-09 — roughly 28 merged lecture PRs here have not flowed through, and bayes_intro, msy_fishery, and tax_smooth have no translation at all. Wiring in the same automation that lecture-python-programming runs will stop the drift permanently.

Todo

1. Reconcile existing drift (before turning automation on)

• Regenerate the onboarding report against current HEADs (translate status / onboarding) — the sample in [action-translation] Sample onboarding report lecture-python-intro -> lecture-intro.zh-cn #676 is from 2025-12 and stale
• Work through the per-file drift decisions (source→target RESYNC vs backport vs manual) from the fresh report
• Fix known target defects surfaced by [action-translation] Sample onboarding report lecture-python-intro -> lecture-intro.zh-cn #676: missing contributors in about.md Credits (Jiarui Zhang, Margaret Beisenbek, Phoebe Grosser) and the cagan_adaptive.md math error (\exp(m_t^d / p_t) should be \exp(m_t^d - p_t))
• Translate the 3 missing lectures (bayes_intro, msy_fishery, tax_smooth) via translate init/forward
• RESYNC the drifted files (translate forward) so target matches current source

2. Wire the automation (this repo)

• Add .github/workflows/sync-translations-zh-cn.yml, copied from lecture-python-programming with target-repo: QuantEcon/lecture-intro.zh-cn (pin QuantEcon/action-translation@v0.15.0; include the \translate-resync comment trigger)
• Confirm TRANSLATION_PAT and ANTHROPIC_API_KEY secrets are set on this repo

3. Review workflow (target repo)

• Land ENH: add review-translations workflow for action-translation lecture-intro.zh-cn#212 (review-translations workflow, open since April)
• Adopt rebase-translations.yml from lecture-python-programming.zh-cn for parity

4. Verify end-to-end

• Trigger one sync (next merged lecture PR, or \translate-resync zh-cn on a recent one) and confirm a translated PR opens, passes AI review, and builds in the target repo
• Tick off Track A in Phase 0 — wire sync automation into the two unwired zh-cn series (tracking) action-translation#74

[mmcky]

Phase 0 Track A is complete (2026-07-18) — closing.

Step	Outcome
Secrets	ANTHROPIC_API_KEY + QUANTECON_SERVICES_PAT org-visible to this repo; ANTHROPIC_API_KEY also extended to the target (its review/rebase workflows need it)
Drift reconciliation	41-lecture translate forward wave — QuantEcon/lecture-intro.zh-cn#225–#265, all merged after structural + localisation review (11 branches hand-fixed to preserve campaign localisation)
Target onboarding	.translate/ config + state + heading maps: QuantEcon/lecture-intro.zh-cn#266; translate doctor green
Wiring	sync #795 (this repo) + review/rebase QuantEcon/lecture-intro.zh-cn#212, all @v0
E2E verification	\translate-resync zh-cn on #793 → QuantEcon/lecture-intro.zh-cn#267 opened, AI review PASS (8.8/10), merged

The 3 missing lectures (bayes_intro, msy_fishery, tax_smooth) are Phase 1 gap-fill, deliberately out of scope here. Engine feedback from the run: QuantEcon/action-translation#104, #105, #106, #107. Full run report in the program workspace (reports/2026-07-18-intro-zh-cn-resync-wave.md).

🤖 Generated with Claude Code