Helps solve: freelawproject/juriscraper#858

- New command to re-run Site.extract_from_text over downloaded opinions
- Able to filter by Docket.court_id ,  OpinionCluster.date_filed, OpinionCluster.precedential_status
- Updates tasks.update_from_document_text to return information for logging purposes
- Updates test_opinion_scraper to get a Site.extract_from_text method

- validate citation objects from `Site.extract_from_text`. Add tests for this
- abstract --courts required argument for scrapers into ScraperCommand class
also, made it more flexible
- refactor cl_scrape_opinions; cl_scrape_oral_arguments to account for this
- delete cl.scrapers.utils.extract_recap_documents which was generating a
circular import. This function was not used anywhere