Implement channel and playlist browsing

BrowsePage accepts pasted or routed channel/playlist URLs:
- Channels get Videos / Shorts / Live sub-tabs mapped to the channel's
  /videos, /shorts and /streams listings, lazily fetched 24 at a time
  with -I range pagination; channel title and id resolve from a cheap
  -I 1:1 metadata fetch. Subscribe emits subscribeRequested for the
  Feed page to wire up.
- Playlists get a single grid with Play all (bulk-enqueues into the
  Watch queue) and a Download button that deep-links the playlist into
  the Downloads tab.

Workers are parented to their pages and fetch errors surface as status
text instead of crashing (verified against a channel with no videos
tab). fetch_flat_info gains an items="1:1" limiter so metadata probes
no longer enumerate whole channels.

Verified live: kurzgesagt channel (24 cards, title+id resolved) and a
17-video playlist with Play-all queueing.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

ytsage/core/ytsage_client.py

@@ -160,9 +160,17 @@ class YtdlpClient:
         """Full info dict for a single video, including formats[].url."""
         return self.run_json(["--dump-single-json", url], timeout=timeout)
 
-    def fetch_flat_info(self, url: str, timeout: int = 300) -> Dict[str, Any]:
-        """Flat info dict (playlist/channel entries without per-video formats)."""
-        return self.run_json(["--dump-single-json", "--flat-playlist", url], timeout=timeout)
+    def fetch_flat_info(self, url: str, timeout: int = 300, items: Optional[str] = None) -> Dict[str, Any]:
+        """Flat info dict (playlist/channel entries without per-video formats).
+
+        items: optional -I range like "1:1" to limit entries when only the
+        top-level metadata (channel title/id) is needed.
+        """
+        args = ["--dump-single-json", "--flat-playlist"]
+        if items:
+            args.extend(["-I", items])
+        args.append(url)
+        return self.run_json(args, timeout=timeout)
 
     def fetch_flat_entries(
         self, url: str, start: int = 1, end: int = 30, timeout: int = 120, use_cache: bool = True