feat: add selective audio stream normalization

[slhck]

Add support for selective audio stream normalization, addressing issue #285.

New CLI arguments:

• -as/--audio-streams: Select specific audio streams by index (comma-separated)
• --audio-default-only: Only normalize audio streams with 'default' disposition
• --keep-other-audio: Keep non-selected streams as passthrough (unchanged)

Implementation details:

• Added is_default flag to AudioStream to track disposition
• Updated MediaFile.parse_streams() to parse "(default)" flag from FFmpeg output
• Added _get_streams_to_normalize() method to filter streams based on selection
• Updated first pass, filter generation, and second pass for selected + passthrough streams
• Proper output stream index tracking for correct codec assignment

Validation:

• Prevents using both -as and --audio-default-only together
• Prevents using both --keep-other-audio and --keep-original-audio together

Tests:

• Added 6 new test cases covering all scenarios
• All 46 tests passing (40 original + 6 new)

Documentation:

• Updated docs/usage/options.md with complete documentation for new options