Add embedded provenance metadata to Whisper-generated VTT files

[dinahhandel]

Add the following fields to the VTT file, in JSON format:

{
"media_id": "druid",
"modification_date": "date created",
"provider": "Stanford University Libraries",
"language": "",
"type": "transcript",
"human_review_level": "None",
"application_name": "Whisper",
"application_version": "version",
"application_parameters": "",
"inference_model": ""
}

Language value should be identified in either the Whisper JSON or identified when the media is sent to Whisper. Application information and model should be available from Whisper.