Migrate translation workflows to @v0, adopt the review template, enable -W

Moves review and rebase off the exact `@v0.16.1` pin onto the floating
`@v0` tag, per the pin policy settled in
QuantEcon/project-translation#9, and aligns the review workflow with the
upstream template in action-translation
docs/user/tutorials/connect-existing.md.

Two substantive changes beyond the pin:

**Concurrency moves from workflow level to job level**, and
`cancel-in-progress` from false to true. The previous comment argued
that job-level `true` was unsafe because an 'automated' label event
would cancel the in-flight review for 'action-translation' and then
skip its own job via the filter, leaving no review at all. That was
reasoned from the GitHub docs and never tested, and production
disproved it: across roughly 15 live opportunities on
lecture-intro.zh-cn the group was entered only after the `if` passed,
the 'automated' event skipped without cancelling, and a review was
posted every time. The stale rationale is replaced rather than left in
place.

**A `permissions` block is added** — v0.17.0's review dedupe deletes
superseded comments, which requires `pull-requests: write`.

The pin was two releases behind, which matters here: v0.18.0 carries
the fix for a review-mode defect where a model response missing a
criterion score became NaN and rendered as an automatic FAIL on
otherwise clean PRs. See QuantEcon/action-translation#102.

Separately, `ci.yml` gains `-W` on the notebook build. This repo's
publish.yml already builds strict, so the PR gate was weaker than the
publish gate — a corrupted sync could merge green and then break the
published site. Coverage is 26 of 26 lectures with a `_toc.yml`
identical to source, so there are no untranslated-lecture references
for the flag to trip on. The strict build is the last line of defence
against the silent-corruption class in
QuantEcon/action-translation#118 and #119, both of which surfaced only
because a downstream repo built with -n -W.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: mmcky

.github/workflows/ci.yml

@@ -60,7 +60,7 @@ jobs:
       - name: Build HTML
         shell: bash -l {0}
         run: |
-          jb build lectures --path-output ./ -n --keep-going
+          jb build lectures --path-output ./ -n -W --keep-going
       - name: Upload Execution Reports
         uses: actions/upload-artifact@v7
         if: failure()

.github/workflows/rebase-translations.yml

@@ -34,7 +34,7 @@ jobs:
 
     steps:
       - name: Rebase open translation PRs
-        uses: QuantEcon/action-translation@v0.16.1
+        uses: QuantEcon/action-translation@v0
         with:
           mode: rebase
           anthropic-api-key: ${{ secrets.ANTHROPIC_API_KEY }}

.github/workflows/review-translations.yml

@@ -1,48 +1,50 @@
 # Review Translations — Quality check on translation PRs
 # When a PR is opened/updated that carries the 'action-translation' label,
 # this workflow runs a quality review and posts a comment.
+#
+# Mirrors the upstream template in action-translation
+# docs/user/tutorials/connect-existing.md — keep it in step with that.
 name: Review Translations
 
 on:
   pull_request:
     types: [opened, synchronize, labeled, reopened]
 
-# Serialise reviews per PR.
-#
-# The sync action creates the PR and then applies its labels in a separate call,
-# so a single sync fires `opened` plus one `labeled` event per label, all within
-# a couple of seconds. Every one of those starts a full review, and the action's
-# "update the existing comment, else create one" logic is a check-then-act with
-# no lock — concurrent runs all observe "no comment yet" and each create one.
-# See QuantEcon/lecture-python-programming.fr#6, which collected two review
-# comments this way, and QuantEcon/action-translation#96 for the upstream bug.
-#
-# cancel-in-progress is deliberately false. The labels are applied in one API
-# call, so event ordering is not guaranteed; if 'automated' arrived last it would
-# cancel the in-flight review for 'action-translation' and then skip its own job
-# via the filter below, leaving no review at all. Queuing instead means the first
-# run creates the comment and any later run updates it — one comment, always.
-concurrency:
-  group: review-translations-${{ github.event.pull_request.number }}
-  cancel-in-progress: false
-
 jobs:
   review:
-    # Require the 'action-translation' label, and — for `labeled` events — ignore
-    # labels other than that one. Without the second clause the 'automated' label
-    # fires a second, redundant review of the identical diff.
-    if: >-
+    # Ignore `labeled` events for every other label: a sync adds its labels in a single
+    # addLabels call, but GitHub emits one `labeled` event per label, and each would
+    # otherwise start a full (billed) review of the same diff.
+    if: >
       contains(github.event.pull_request.labels.*.name, 'action-translation') &&
-      (github.event.action != 'labeled' ||
-       github.event.label.name == 'action-translation')
+      (github.event.action != 'labeled' || github.event.label.name == 'action-translation')
     runs-on: ubuntu-latest
 
+    # v0.17.0's review dedupe deletes superseded comments, which needs pull-requests: write.
+    permissions:
+      contents: read
+      pull-requests: write
+
+    # One review per PR — supersede an in-flight review instead of running both.
+    # Job-level (not workflow-level) on purpose: the group is entered only after the `if`
+    # above has passed, so a `labeled` event for 'automated' skips out without cancelling
+    # the real review. At workflow level it would cancel first and skip second, leaving none.
+    #
+    # This supersedes the workflow-level `cancel-in-progress: false` previously used here,
+    # whose rationale — that an 'automated' label event would cancel the real review and
+    # then skip its own job, leaving none — was reasoned from the docs and never tested.
+    # Production disproved it: across ~15 live opportunities on lecture-intro.zh-cn the
+    # 'automated' event skipped without cancelling and a review was posted every time.
+    concurrency:
+      group: review-translations-${{ github.event.pull_request.number }}
+      cancel-in-progress: true
+
     steps:
       - uses: actions/checkout@v7
         with:
           fetch-depth: 2
 
-      - uses: QuantEcon/action-translation@v0.16.1
+      - uses: QuantEcon/action-translation@v0
         with:
           mode: review
           source-repo: QuantEcon/lecture-python-programming