onion service scanning: check meta tags for onion-location

[redshiftzero]

Closes #269. Followup to #262.

This PR:

• Changes onion-location to a nullable field. This is motivated primarily by wanting to have a way to distinguish between failed and successful scans, both for anyone who were to consume this data but also for the STN Twitter bot (that code is here).
• Gets the page content using requests, parses the page content using lxml, then sees if there is an onion-location meta tag

The easiest way to test once on this branch is to:

1. Ensure you have the test data added (so you have some sites to test on). Then run a scan: make dev-scan
2. Visit the API: http://127.0.0.1:8000/api/v1/sites/
3. For any sites that failed during the scan, they should have a null onion_available. Otherwise, Propublica should have onion_available as true, and techcrunch should have onion_available as false.

[redshiftzero]

thanks for reviewing! the last commit (e8ec6bc0a696695ca5935fcc5aeff067aceb9da7) is the new change here, the prior commits are just rebasing on the latest

[chigby]

@redshiftzero This latest change looks great. I can merge once this is up-to-date with the base branch.