Replace supervisord with further systemd intergration

[phillxnet]

At one time, before systemd, we did all our service orchestration via supervirord: we have progressively moved to using systemd services for our various 'parts' but have yet to migrate the following:

• gunicorn (/opt/rockstor/.venv/bin/gunicorn) - runs our main wsgi:application.
• replication daemon (/opt/rockstor/.venv/bin/replicad)
• data-collector (/opt/rockstor/.venv/bin/data-collector) - live-info
• Huey (/opt/rockstor/.venv/bin/django-admin run_huey ...) longer running tasks manager.

It is proposed that these be incorporated into rockstor.service via multiple ExecStart entries. Or possibly host only gunicorn in rockstor.service and create dedicated services for each of the other services. This would normalise all our service management on a single orchestration mechanism and one native to the base OK. It would also remove a central Python dependency.

There may however be concurrency/threading issues that are otherwise handled better by supervisord given it is itself, a python app.