Fix CI errors during systemtests caused by SchedulerDB caching

As seen in our CI, it's common for the late staged of our CI to fail in some of the system tests:

• https://gitlab.cern.ch/cta/CTA/-/pipelines

These errors are random, and usually disappear after retrying a few times.

However, this becomes a waste of the CI resources and may be hiding some real issue in our code base.
We should aim to fix it.