fix(pixiv): capture ugoira frame timings in the post record (ordering bug)

The core writes the post record BEFORE extract_media, but the ugoira frame
delays were only memoized DURING extract_media — so write_post_record never saw
them and ugoira_frames was always empty in the record. Extract a memoized
_ugoira_meta (frames + zip url share ONE /v1/ugoira/metadata call regardless of
order) and inject client.fetch_ugoira_frames into the downloader (mirrors
Patreon's content_fetcher) so write_post_record populates the frames itself.
Zero extra API calls — the fetch is shared/memoized with extract_media. A
recapture now backfills the timings onto existing ugoira posts.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM

tests/test_pixiv_downloader.py

@@ -268,6 +268,40 @@ def test_write_post_record_includes_ugoira_frames(tmp_path):
     assert data["ugoira_frames"] == frames
 
 
+def test_write_post_record_fetches_ugoira_frames_when_absent(tmp_path):
+    # The core writes the post record BEFORE extract_media, so frames aren't
+    # memoized yet — the injected fetcher must populate them so the record keeps
+    # the timings (regression: they were silently always empty).
+    post = _post_only(333)
+    assert "_ugoira_frames" not in post["_work"]
+    frames = [{"file": "000000.jpg", "delay": 120}]
+    calls = []
+
+    def fetcher(p):
+        calls.append(p)
+        p["_work"]["_ugoira_frames"] = frames
+
+    dl = PixivDownloader(
+        tmp_path, validate=False, session=_FakeSession(),
+        ugoira_frames_fetcher=fetcher,
+    )
+    outcome = dl.write_post_record(post, "artist-a")
+    data = json.loads(outcome.path.read_text())
+    assert data["ugoira_frames"] == frames
+    assert len(calls) == 1
+
+
+def test_write_post_record_non_ugoira_does_not_call_fetcher(tmp_path):
+    post = _post_only(111)                       # a plain multi-page illust
+    calls = []
+    dl = PixivDownloader(
+        tmp_path, validate=False, session=_FakeSession(),
+        ugoira_frames_fetcher=lambda p: calls.append(p),
+    )
+    dl.write_post_record(post, "artist-a")
+    assert calls == []
+
+
 def test_write_post_record_without_id(tmp_path):
     dl = _downloader(tmp_path)
     outcome = dl.write_post_record({"id": None, "attributes": {}}, "artist-a")