feat: Add channel alias support, debug export, and improve tag handling

[RedShieldArr]

feat: add alias-aware matching, debug export, and ignored-tag expansion

Improve channel matching quality and provide deeper debug visibility.

• Add premium channel alias support

  • Load aliases from country JSON databases
  • Match against aliases and map back to canonical channel names
  • Expand UKTV alias coverage in UK channel data

• Add Debug Match Export action

  • Export detailed matching diagnostics to CSV
  • Include OTA detection, normalized queries, and fuzzy match stages
  • Add setting to control top N fuzzy candidates in debug output

• Improve ignored tag handling

  • Add shared ignored-tag expansion helper
  • Support bracketed, parenthesized, and bare tag variants from one entry
  • Apply consistently across rename and category workflows

• Improve Unicode matching behavior

  • Preserve accented names during normalization for better matching
  • Keep accent handling in the dedicated normalization pipeline
  • Correct related normalization comment text

• Update documentation

  • Document Debug Match Export action and debug-top-N setting
  • Clarify ignored-tag auto-expansion behavior
  • Expand CSV export format section with debug export details

[PiratesIRC]

Thanks for this — the alias support, the _expand_ignored_tags refactor, and the debug export are all genuinely useful. A few things need addressing before this can land:

Blocking: branch is based on the pre-ORM codebase

This PR predates the HTTP-API → Django ORM migration. main is now v1.26.1001200 and accesses data via from apps.channels.models import .... The new debug_export_action calls self._get_api_token(...) and self._get_api_data("/api/channels/groups/", ...), which no longer exist on main (replaced by _get_all_groups / _get_all_channels ORM helpers). The run() dispatch and the category_groups_dry_run_action / organize_by_category_action call sites you patched were also rewritten for ORM — hence the merge conflicts. Could you rebase onto current main and re-apply the concepts (not the diff)?

Perf: fuzzy_match_debug bypasses the token pre-filter

It iterates all candidate_names three times (stages 1–3), recomputing normalize_name each pass, with no get_candidates() pre-filter and no normalization cache. That reintroduces the O(streams×channels) cost the v1.26 token-index work removed (32h → 6s). Over ~31K channels a debug export could hang. Please route it through get_candidates() + _get_cached_norm() like fuzzy_match does.

Behavior change in tag expansion

_expand_ignored_tags now also appends the bare inner token (the old inline blocks only added the opposite-bracket form). Bare HD/4K stripping is more aggressive and could over-strip names where the token appears as a word. It's documented in the README, but please spot-check matching quality on a real channel list.

PR description vs. diff

The description mentions Unicode/accent-preservation normalization changes and comment fixes, but no such changes appear in the diff (the fuzzy_matcher hunks are only __init__, db-load, and the new debug method). Could you reconcile the description with what's actually included?

Minor

• import json inside debug_export_action is unused (the module already imports it).
• UK_channels.json has pre-existing duplicate channel_name entries (U&Eden HD, U&Yesterday HD, U&Yesterday+1 each appear twice); this PR duplicates the alias blocks into both, with inconsistent key ordering. Harmless at runtime (dict last-wins), but the dedup is worth splitting into its own change.

The alias→canonical reverse mapping itself is sound, and remapping before the category_map_premium lookup is correctly ordered. Happy to re-review once it's rebased onto ORM main.