Onto API scaling update

[colinkcurtis]

• [ ] Use gunicorn, with N workers, to launch the Onto API
• [ ] Dockerize the new gunicorn variant of Onto API
• [ ] Re-launch the Onto API application with this new schema
• [x] rename 'ancestors' to 'superterms'
• [x] rename 'descendants' to 'subterms'
• [x] convert 'descendants' endpoint to a recursive search for all (to the bottom of the tree) 'is_a' descendants of an input CURIE
• [x] restore the 'is_a' ancestry check endpoint
• [x] enable the 'single_level_is_a' endpoint for finding DIRECT, single generation down, 'is_a' descendants

Estimated Completion: Postponed in favor of UberOnto API implementation.

[colinkcurtis]

update: The gunicorn variant of the Onto API is now working locally. Looking to have it dockerized and working behind onto.renci.org/apidocs by Wednesday, February 27.

[colinkcurtis]

Tangentially related: the gunicorn-ificiation of this service was brought up to improve stability of the service. That is still in process. As an attempt at a shorter-term fix, I created a new Marathon/Mesos application using the same set of Onto API files and docker containers and configurations. This has increased the stability of the application for unknown reasons. A temporary improvement.