DOC: show adaptive algorithm performance

[stsievert]

What does this PR implement?
This cleans some experiments, and moves them to a separate repo. It will be open until those experiments are complete.

It also finally gets adaptive algorithms working, and integrates it into the docs.

TODO:

• removes (unused) experiment on number of users.
• implement new plotting on dashboard
• post n_queries_scored_before_limiting to database for RR.
• add worker rotation in the submit functions on backend.
• resolve posterior issues (reversed?). This is based off the removed experiment, but in a new (unpublished) repo.

Reference issues/PRs