BUG - Client connection timeout stops indexing

[justincorrigible]

Describe the bug

As noticed in VirusSeq, when a request is made for Maestro to index a study (e.g., from its swagger, or using postman), if the connection times out, Maestro stops indexing without notification.
In our case, as a workaround, we modified Nginx's ingress annotation for read timeout; which is not ideal and got the job done, but doesn't solve the underlying issue.

Expected behaviour

If Maestro will stop indexing for whatever reason, a log message should be produced to indicate the process has been stopped on purpose. Alternatively, rather than stopping the process when the connection drops (which could happen for a myriad of unexpected reasons), Maestro could immediately respond with a "201. Starting indexing study XYZ", and it should then continue indexing in the background, at least until a command to stop or cancel is sent.

Furthermore, If it doesn't already exist, another endpoint may be implemented for users to inspect which studies are in progress etc.

---

Edit: linking icgc-argo/roadmap#874 to distinguish them, for even though is similar enough, they're not quite the same problem.