BF-2 reboots and upgrades need to be handled properly in the two-cluster design.

Flow of dpu lifecycle controller:

1. Creating deployment with pod that only run sleep
2. Creating pdb that will block pod draining
3. In case tenant kubeconfig is not set allow user to remove pdb manually if host is unschedulable.
4. In case tenant kubeconfig is set and node mapping exists create nodemaintenance cr on tenant cluster
5. On node uncordon delete NM cr and create pdb as it was