Fix automation scenarion maint notification

[kiryazovi-redis]

The current scenario tests have 3 issues which we have to fix for our nightly CI to work.

1. they would not work so well with a cluster that has multiple dbs, so I fixed the logic to choose:
   Master node of our target db
   Any node without shards of the target db
2. They require more logs (This part of the PR can be removed, depending on comments)
3. They take too long due to clean-up. The clean-up can be mostly removed besides the failover one. Because migrate and bind are operations that should take the cluster into usable for next scenario tests. The only exception is migrate only tests.
   (a notification is sent only when we did mgirate + bind, because the proxy is pointing towards "empty" node)

[petyaslavova]

LGTM.