Fluentd flush logs when Elasticsearch is not healthy overwhelming the logstore

Description of problem:
Fluentd continue to push logs even if the Elasticsearch cluster is unhealthy

Version-Release number of selected component (if applicable):
cluster-logging.5.2.2-21

How reproducible:
100%

Steps to Reproduce:
1. Deploy a default Clusterlogging instance
2. Delete the ES pods "oc delete pod -l component=elasticsearch"
3. Verify on the fluentd logs that the log collector component was flushing while the ES cluster wasn't ready.

Actual results:
In the customer environment, this caused ES to never start properly because overwhelmed by logs. Until they completely stoped all Fluentd instances.

Expected results:
Fluentd pods should not push data until ES is healthy.

Additional info:
A workaround exists and is to stop all the Fluentd instances to allow the ES cluster to start properly.