This epic is about adding support for the cluster autoscaler command line flag "--cordon-node-before-terminating". This flag will allow users to configure the autoscaler to either cordon, or not cordon, a node before terminating. This behavior can affect on how long nodes remain in the cluster and whether workloads should be terminated early when deleting nodes.

The command line flag exists in the cluster autoscaler, to complete this epic we need to do the following:

1. add a field to ClusterAutoscaler resource to allow enabling or disabling the cordon node before terminating behavior
2. add logic to the cluster autoscaler operator to read the new field and add the command line flag to the autoscaler deployment
3. write unit tests to confirm the flag enable/disable behavior
4. write end-to-end tests to confirm the behavior when enabling or disabling the flag
5. update documentation to mention new field and behavior