- Implemented audio concatenation using ffmpeg for improved performance
and reliability in _combine_audio_objects.
- Added error handling to fall back on Opus re-encoding if stream-copy
fails.
- Updated _persist_composite_artifacts to calculate audio duration from
page narrations.
- Introduced unit tests to validate ffmpeg integration and error
handling scenarios.