replace k8s wait job with CH based decision

[srikanthccv]

The hooks have been painful to understand or work with. Replace the k8s wait job with some status marker in the ClickHouse schema migrator table flag.

[wSedlacek]

My team has observed this as well.
With versions prior to 0.55.0 the hooks were flaky and seemed to have conditions where a job wouldn't be created when it was expect to, or the pod would take too long to place and the job would be gone by the time to pod placed.

On 0.55.0 I am having trouble even installing because durning setup Zookeeper cycles causing EBS to need to remount all the drives on different nodes (multi attach volume error until it remounts) and by the time Zookeeper remounts the -sync job has already failed 5 times and given up.

I was only able to get lucky once after 10 tries where all the zookeeper pods placed on the exact same nodes and there was no delay with shuffling volumes.

(Note: using efs instead of ebs would have also worked but this cluster is not configured with the efs controller)