fix: close remaining concurrent yt-dlp session sources

Three more code paths were bypassing the _meta_lock guard and firing
raw yt-dlp processes concurrently with active downloads:

- Popular fetch Phase 1 (flat-playlist channel crawl): changed from
  ytdlp._run to ytdlp._meta_run so it waits for active downloads
- download_subs_only: changed from _run to _meta_run
- fetch_video_comments: returns empty list immediately if a download
  is active (avoids blocking a 90s call indefinitely)
- Diagnostic test endpoint (settings): switched to _meta_run

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

backend/services/ytdlp.py

@@ -607,7 +607,7 @@ def download_subs_only(video_id: str, subtitle_langs: str) -> bool:
     """Download subtitle files only (no video) for an already-downloaded video."""
     url = f"https://www.youtube.com/watch?v={video_id}"
     output_template = str(Path(settings.download_path) / f"{video_id}.%(ext)s")
-    _, _, code = _run([
+    _, _, code = _meta_run([
         "yt-dlp", url,
         "--skip-download", "--no-playlist",
         "--write-subs", "--write-auto-subs",
@@ -656,6 +656,11 @@ def fetch_video_comments(youtube_video_id: str, max_comments: int = 20) -> list[
     import os
     import tempfile
 
+    # Don't fire a concurrent yt-dlp session while a download is running — it
+    # causes YouTube to see two simultaneous authenticated sessions and invalidates cookies.
+    if is_download_active():
+        return []
+
     url = f"https://www.youtube.com/watch?v={youtube_video_id}"
 
     with tempfile.TemporaryDirectory() as tmpdir: