Enable real chunks job

[l0kix2]

Fix #396

I've added k8s job in the master component. The job simply checks if ytsaurus image version is between 23.2—24.1 including and if yes — sets //sys/@config/node_tracker/enable_real_chunk_locations=%true according to the issue.
It is being executed:

• when the cluster is Initilazing (after master pods are created and before dnd pods are created)
• on each master update (in FullUpdate mode since we want dnds to be restarted also) after master in safe mode but before all pods are recreated (because we want flag before dnd restart).

This is required to prepare user clusters for 24.2. We will ask users to update their cluster to some (23.2—24.1) version with ytop version with this code prior to 24.2 update.

How it was tested

1. I've added that flag is set to true after update for current versions we test against (create 23.2.0 and update to 23.2.1 + create 23.2.0 and update to 24.1.0) + no node with use_imaginary_chunk_locations=true exist after the update.
2. I've manually tested that flag is not set for 23.1 (we don't test against it, kinda unsuported i'd say)
3. I've manually tested that update from 23.1 (flag false) to 23.2 sets flag to true.

Other

We also may implement some check for possibility of update to 24.2 checking data nodes meta information in a separate issue.

[achulkov2]

All contents LGTM, but I think it is useful to add a validation job that would stop the update to 24.2 if it would lead to master cores.