Implement new crawler based on wpull

[chosak]

This PR adds an alternate method of crawling a website based on wpull.

The current approach uses 2 steps:

1. Use wget to crawl a website, generating a WARC file
2. Run a Django management command (warc_to_db) to convert the WARC to a queryable SQLite database

The new approach uses only a single step:

1. Use wpull plus a custom plugin to crawl a website directly into a queryable SQLite database

This can be done using a new Django management command:

% ./manage.py crawl --help
Usage: manage.py crawl [OPTIONS] START_URL DB_FILENAME

  Crawl a website to a SQLite database.

Options:
  --max-pages INTEGER            Maximum number of pages to crawl
  --depth INTEGER                Maximum crawl depth
  --recreate                     Overwrite SQLite database if it already
                                 exists  [default: False]
  --resume

Because wpull unfortunately doesn't support Python greater than 3.6 (ArchiveTeam/wpull#426), this new approach requires downgrading the runtime of this repo to Python 3.6 as well. This in turn requires downgrading Django from version 4.x back to 3.2.