Fix taped coredumping when Postgres scheduler enabled

After #1423 , it seems that this was caused by a test-repack job in a Postgres pipeline: https://gitlab.cern.ch/cta/CTA/-/pipelines/13491078

Trying to run test_repack.sh locally with the Postgres scheduler enabled produces a significant number of core dumps (most likely the reason behind the machine filling up).

Will investigate and do a fix.