Add timeouts for cluster metadata initialization and for init containers

[lhotari]

Motivation

• Sometimes the Pulsar deployment doesn't complete, and the broker services don't become available. When investigating the problem, it seems that cluster metadata initialization might get stuck and that blocks starting the pods.
• This problem seems to happen with Pulsar 2.8.x+ and Zookeeper, when TLS is enabled for Zookeeper. #190 is the PR to switch to use Pulsar 2.8.x .

Modifications

• Add timeouts for waiting for zk and bk to become available.
  • timeouts will help failures recover eventually
• Add timeout to cluster metadata initialization jobs