API/DOC/MAINT: update for experimental use

[stsievert]

What does this PR implement?
I am in the process of fine-tuning Salmon performance to match NEXT's performance. So, that means I am explicitly mirror a NEXT experiment in salmon-experiments.

This PR makes the following changes:

• Adds new alien-eggs dataset, and updates the noise model to mirror the NEXT paper.
• Makes some changes around resetting (in hopes of solving Visiting /reset doesn't restart adaptive algorithms #82)
• Does not post queries with NaN score to the database.
• Better chooses max_epochs for adaptive runners.
• Refreshes documentation (better + more complete API docs, small fixes)
• Updates documentation with experiment that changes the response rate of Salmon.

Reference issues/PRs
#82, resetting.