Description of problem:

In the control plane machine set operator we perform e2e periodic tests that check the ability to do a rolling update of an entire OCP control plane.

This is a quite involved test as we need to drain and replace all the master machines/nodes, altering operators, waiting for machines to come up + bootstrap and nodes to drain and move their workloads to others while respecting PDBs, and etcd quorum.

As such we need to make sure we are robust to transient issues, occasionaly slow-downs and network errors.

We have investigated these timeout issues and identified some common culprits that we want to address, see: https://redhat-internal.slack.com/archives/GE2HQ9QP4/p1678966522151799