Updating cluster should have higher timeout on waiting for green status

Description

During updating cluster (bumping Elasticsearch version) shard allocation is disabled, then when cluster with new version starts with yellow status shard allocation is enabled again. There cluster marks shards on each index, this process is time consuming and the more indices exists the more time is needed.

Proposal

Read how many indices exists on cluster and rise time to wait for green status if the number of indices is large enough.

For example number_of_indices/2 seconds

Proposal 2

Add optional variable which will tell role how many retries it should make before it will fail on check_status action.