[backport-1.6]: multus-cleanup scheduled in loop when node is under disk-pressure

A solution for issue #3244 (closed) needs to be backported to 1.6.x