see: https://github.com/metabrainz/picard/pull/2709#issuecomment-3239928988

and see: https://github.com/metabrainz/picard/pull/2709#issuecomment-3240021428

The existing `columns.py` contains several types of columns:

1. *Simple Field Columns*: `Column(N_("Artist"), 'artist')`
2. *Default Columns*: `DefaultColumn(N_("Title"), 'title', ...)`
3. *Custom Sort Columns*: Using `sortkey` functions like `_sortkey_length`, `_sortkey_filesize`
4. *Special Columns*: `MatchQualityColumn`, `IconColumn` (fingerprint status)

Benefits of Refactoring

1. *Code Unification: Eliminates duplicate sorting logic across multiple `sortkey` functions
2. *Consistency*: All columns use the same provider-based architecture
3. *Maintainability*: Centralized sorting logic in reusable adapters
4. *Extensibility*: Easy to add new sorting behaviors without creating new functions
5. *Type Safety*: Better type hints and validation through the provider system

Implementation Plan

Phase 1: Create Missing Providers
1. *MatchQualityProvider*: Implement custom provider for match quality column
2. *Enhanced NumericSortAdapter: Handle edge cases from current `sortkey` functions
3. *StatusIconProvider*: For columns with status icons

Phase 2: Refactor Standard Columns
1. *Simple columns*: Replace `Column()` with `make_provider_column()`
2. *Numeric columns*: Use `NumericSortAdapter` with `FieldReferenceProvider`
3. *Natural sort columns*: Use `NaturalSortAdapter`
4. *Special columns*: Create custom providers

Phase 3: Cleanup
1. *Remove redundant functions: Delete `sortkey` functions
2. *Update imports*: Remove unused imports
3. *Testing*: Ensure all sorting behaviors are preserved

Migration Strategy

*Backward Compatibility*: The refactoring can be done incrementally:
1. Keep existing `Column` classes working
2. Gradually replace with `make_provider_column()` calls
3. Remove old code only after verification

*Testing Approach*:
1. Create comprehensive tests for each sorting adapter
2. Verify sorting behavior matches current implementation
3. Test edge cases (empty values, invalid data, etc.)

Challenges & Solutions

1. *MatchQualityColumn Complexity*: 
   - *Solution*: Create dedicated `MatchQualityProvider` that handles both display and sorting

2. *Status Icon Integration*:
   - *Solution*: Extend `FieldReferenceProvider` or create `StatusIconProvider`

3. *DefaultColumn vs Column*:
   - *Solution*: Use `always_visible=True` parameter in `make_provider_column()`

4. *ImageColumn Special Cases*:
   - *Solution*: Keep `ImageColumn` for complex rendering, use providers for simple cases