Import audio/video files through the transcription -> summary pipeline

[wynnwu]

Let users drop in audio/video files that run through transcription (whisper) and then the normal summarize pipeline (specs/002 FR-017/018). Depends on the transcription backend from the recording work.