feat: add content_path filter for precise torrent deletion

[sysit]

Summary

Add support for filtering torrents by their download directory (content_path), allowing users to apply different retention policies based on where torrents are downloaded.

Motivation

Currently, users can filter torrents by tracker, category, and status. However, there's no way to differentiate torrents based on their download location. This feature enables more precise cleanup strategies, e.g.:

• Delete old movies after 30 days, but keep TV shows longer
• Apply different policies to different storage volumes
• Protect important directories from automatic cleanup

Changes

New Files

• autoremovetorrents/filter/contentpath.py - ContentPathFilter implementation

Modified Files

• autoremovetorrents/client/qbittorrent.py - Fetch content_path from qBittorrent API
• autoremovetorrents/client/transmission.py - Fetch downloadDir from Transmission API
• autoremovetorrents/strategy.py - Register ContentPathFilter
• autoremovetorrents/torrent.py - Document content_path attribute

Usage Example

my_task:
  client: transmission
  host: http://127.0.0.1:9091
  username: admin
  password: password
  strategies:
    # Delete movies after 30 days
    movies_cleanup:
      content_paths:
        - /downloads/movies
      remove: seeding_time > 2592000 or ratio > 2
    
    # Delete TV shows after 14 days, but protect important ones
    tv_cleanup:
      content_paths:
        - /downloads/tv
      excluded_content_paths:
        - /downloads/tv/important
      remove: seeding_time > 1209600

  delete_data: false

Features

• ✅ Exact path match
• ✅ Parent directory match (/downloads/tv matches /downloads/tv/show)
• ✅ Multiple content paths support
• ✅ excluded_content_paths for fine-grained control
• ✅ Works with both qBittorrent and Transmission clients

Testing

Tested with:

• qBittorrent 4.x
• Transmission 3.00
• 1400+ torrents, filtered by content_path + tracker + seeding_time

Checklist

• Code follows project style
• Backwards compatible (new optional filter)
• Tested with real Transmission instance