Lock not released when "insufficient memory" for pod

[ppatierno]

When the resources memory "request" cannot be handled due to insufficient memory, the lock is not released and the cluster operator stuck.
Scenario ...
Create a cluster with default configuration.
Update the cluster config map adding a kafka-resources with a memory request too high for the current available memory.
The cluster operator starts rolling update, shuts down the first kafka broker and tries to spin up a new one but "0/1 nodes are available: 1 Insufficient memory." happens.
The lock is never released. The cluster operator is unusable.

I guess it should happen for Zookeeper as well.

[scholzj]

This looks like an issue caused by missing timeout in rolling update. In such case this will be a duplicate for #127