Rework alignment logic to use API

[glynnbird]

Currently the librosa-powered alignment tech calculates the "offset" for each non-leader part and uses it to trim the videos, which make it to the final rendered output. The offset for each part isn't recorded anywhere and isn't tweakable.

• ensure that when a librosa-powered alignment operation is undertaken, it's calculated offset is stored using the API (see below)
• the JS renderer will take the "offset" values for each songpart from the database and save them in the render config JSON
• the Python renderer will respect the offsets it sees in the JSON file
• the trimmed bucket will be decomissioned
• if we choose to tweak an offset in the database (or via a future front-end update) then the results would be reflected in the JSON file and the rendered video.

To record the offset via the API, the following API call is required:

curl -X POST -d'{"choirId":"x", "songId":"y","partId":"z","offset":100}' -H 'Content-type: application/json' "$URL/choir/songpart"

See

• https://github.com/Choirless/choirlessapi/blob/master/API.md#post-choirsongpart
• https://github.com/orgs/Choirless/projects/3#card-43852148