fix(sync): try a channel's feed again before counting it as failed

YouTube's feed endpoint answers the same channel with 404, 500 and 200 in
turn: on prod 182, then 296 of 396 channels failed a poll, and a probe of
24 channels saw 15 fail a first request while 16 had succeeded by the
fifth. A channel that failed once waited for the next poll, 20 minutes
later, so new uploads showed up late.

A feed is now tried up to three times (after 1 s and 2 s) on a transport
error, a 404, a 429 or a 5xx; a refusal such as 403 is not retried. The
last failure is what gets counted. A feed that is really gone costs two
more requests per poll.

backend/app/youtube/rss.py

@@ -1,6 +1,7 @@
 """Free per-channel RSS feed reader (no API quota). Returns the channel's most recent
 ~15 uploads as lightweight stubs for fast fresh-video detection."""
 
+import time
 from datetime import datetime, timezone
 
 import feedparser
@@ -8,6 +9,15 @@ import httpx
 
 RSS_URL = "https://www.youtube.com/feeds/videos.xml?channel_id={channel_id}"
 
+# The waits before a second and a third try. YouTube's feed endpoint answers the SAME channel with
+# 404, 500 and 200 in turn (measured 2026-09-19: 15 of 24 channels failed a first request, 16 of them
+# had succeeded by the fifth, one second apart), so one failure is not an answer. A feed that is
+# really gone costs two more requests per poll.
+RETRY_DELAYS = (1.0, 2.0)
+# A 404 is in the set because that is how the flaky endpoint fails most often; a 403/410 is not,
+# and neither is anything else that says the request itself is wrong.
+_RETRYABLE_STATUS = frozenset({404, 429, 500, 502, 503, 504})
+
 
 class RssError(Exception):
     """The feed could not be read (transport failure or a non-200 answer).
@@ -24,21 +34,34 @@ class RssError(Exception):
         self.reason = reason
 
 
-def fetch_channel_feed(channel_id: str) -> list[dict]:
-    """The channel's recent uploads as stubs. `[]` means the feed really is empty; a failure
-    raises RssError so the caller can leave the channel un-stamped and log it."""
+def _get_feed(channel_id: str) -> bytes:
+    """The feed's body, after up to `1 + len(RETRY_DELAYS)` tries; the LAST failure is raised."""
     url = RSS_URL.format(channel_id=channel_id)
-    try:
-        resp = httpx.get(url, timeout=20.0, headers={"User-Agent": "Siftlode/1.0"})
-    except httpx.HTTPError as exc:
-        raise RssError(f"RSS fetch failed for {channel_id}: {exc}", "transport") from exc
-    if resp.status_code != 200:
-        raise RssError(
+    for attempt in range(len(RETRY_DELAYS) + 1):
+        if attempt:
+            time.sleep(RETRY_DELAYS[attempt - 1])
+        try:
+            resp = httpx.get(url, timeout=20.0, headers={"User-Agent": "Siftlode/1.0"})
+        # silent-ok: not silent — the last failure is raised after the final try.
+        except httpx.HTTPError as exc:
+            error = RssError(f"RSS fetch failed for {channel_id}: {exc}", "transport")
+            error.__cause__ = exc
+            continue
+        if resp.status_code == 200:
+            return resp.content
+        error = RssError(
             f"RSS fetch for {channel_id} returned HTTP {resp.status_code}",
             f"http_{resp.status_code}",
         )
+        if resp.status_code not in _RETRYABLE_STATUS:
+            break
+    raise error
 
-    parsed = feedparser.parse(resp.content)
+
+def fetch_channel_feed(channel_id: str) -> list[dict]:
+    """The channel's recent uploads as stubs. `[]` means the feed really is empty; a failure
+    (after the retries) raises RssError so the caller can leave the channel un-stamped and log it."""
+    parsed = feedparser.parse(_get_feed(channel_id))
     out: list[dict] = []
     for entry in parsed.entries:
         video_id = entry.get("yt_videoid")