Feature: sync identified sources to database

[josh-chamberlain]

Context

Most of our issues have concerned feeding the machine learning pipeline; this one is about updating our database at the end of all the the flowcharts in the readme.

Requirements

• create a manually-triggered GitHub Action to:
  • check Hugging Face for new sources (they will be synced from label studio: automated workflow: update training data from label studio #89)
  • generate name, description: use NLP model to generate name and description for data sources #43
  • generate agency_described: use agency_identifier in this repo; may need refinement
  • submit them to the database
    • until v2: new data source data-sources-app#252 is done, we can just generate a CSV to upload every week or two
    • we could also use the airtable API
    • use record_type from the labeling process