Convert /api/galaxy endpoints to Fetch requests against Galaxy's web API

[johnwduggan]

Link to Original Issue

#72

Summary of Changes

Polling Galaxy via bioblend has proven inefficient and error-prone. I have rewritten the front-end to poll Galaxy via web transactions since we have already converted it to run it on a subdomain of Galaxy. This reduces the latency of the monitoring loop from 1-5 seconds down to a few dozen milliseconds.

I have additionally added support for listening to server-side events in most cases to detect when job status changes in Galaxy so we don't have to explicitly poll, though some things (such as file ingestion for auto-launch) don't fire SSE events so I have retained a backup polling interval.

The is_admin check still uses the dashboard's server-side since we don't want users to be able to bypass it. We can remove this in the future once we remove the database holding banner messages.

I've tested each non-TODO row in the run sheet in the tests on ndip-dev which currently has these changes deployed. I encourage you to do the same as this is a very fundamental change to the way the dashboard works and regressions will be easy to introduce here.

Checklist

• The PR has a clear and concise title
• Code is self-documented and follows style guidelines.
• Automated tests are written and pass successfully.
• Regression tests (e.g. manually triggered system tests, manual GUI/tool tests, ...) are performed to make sure the PR does not break anything (when applicable)
• Readme file is present and up-to-date.

Documentation Updates

Additional Notes

Closes #72
Closes #75 since we don't need bioblend any more (apart from is_admin which needs to be run server-side and will be removed in a future PR).