[Webservice] Monitor Slurm jobs in separate process

Description

Now that we have a process supervisor for the webservice & serve_overview process, it makes sense for this to be a separate process as well, rather than a thread in the webservice process (which was convenient when we launched it manually). The diff here is mostly just moving code that already exists into a separate file for clarity.

This means its logs will be visible separately, and the supervisor can restart the job monitor if it fails. The overview server (http://max-exfl016.desy.de:8008/ ) will no longer show log messages from job monitoring, which it currently does. We could add them as a separate block if needed, or use caldeploy logs to look at them.

A related change will be needed in the deployment tools.

How Has This Been Tested?

Run on max-exfl017, see comment below.

Types of changes

• Refactor (refactoring code with no functionality changes)

Checklist:

• My code follows the code style of this project.

Reviewers

@schmidtp @roscar

webservice/config/serve_overview.yaml

@@ -12,6 +12,7 @@ shell-commands:
   total-jobs: "sinfo -p exfel -o %A --noheader"
   tail-log: "tail -5000 web.log"
   cat-log: "cat web.log"
+  tail-log-monitor: "tail -5000 monitor.log"
 
 run-candidates:
     - "--run-high"