Some SLRP tests fail under stress

Some tests using storage local resource providers can fail under stress, e.g., on a 72 core machine triggered by running stress-ng --cpu=100 --io 20 --vm 20 --fork 100 --timeout  3600s.

These tests typically attempt to run with paused clock, but have do unpause the clock to allow the CSI plugin container to connect. The framework started as part of these tests can then run into a situation where they never receive any offers.

We should investigate what is happening here, and fix tests making use of the problematic pattern.