Document the disruption risks for using NoExecute taint #9
Description
There is no stopping from a user to configure a node-readiness rule with a NoExecute taint. This taint is strongly enforced in Kubernetes by evicting the exiting workloads on the node that do not carry the tolerations.
When a critical component goes down or becomes unready in 'continuous' evaluation mode, the node readiness controller will add the managed taint on the node. This may be undesirable to evict the existing workloads with 'NoExecute' in some use-cases.
It is good to capture this 'misconfiguration' risk in the project documentation. And explore whether a warning could be added in rule creation user flow.