Make a better initial cluster state

[Krock21]

The operator does not use init_cluster.py script, which leads to poorly configured cluster from the beginning

We need to analyse what is happening in that script and verify that operator does everything that is present there(and relevant)

One important part is configuring ACLs for cypress nodes. I created an issue for //sys nodes.

In this ticket I suggest covering everything else, like schemas, tmp, pool trees, accounts, users

[Krock21]

One more error that we found is caused by absence of //tmp/yt_wrapper/table_storage

Python SDK creates this directory inside a transaction, if needed, creating a bottleneck and leading to these errors:

yt.wrapper.errors.YtCypressTransactionLockConflict: Cannot take lock for child "table_storage" of node //tmp/yt_wrapper since this child is locked by concurrent transaction abcd-efgh

init_cluster.py creates this folder to avoid it. ytsaurus-k8s-operator does not