Reduce memory usage significantly when pulling large channels all at once by
- Not loading entry script until it is being processed
- Deleting the entry memory once fully processed
- Disable yt-dlp from keeping entry info in memory since ytdl-sub reads it from file