Direct communication between Elasticsearch pods in OCP 4

Issue

Elasticsearch is designed specifically to not need a load balancer but all Elasticsearch instances in openshift-logging project communicate through OCP service. Also Kibana and fluentd connects to the Elasticsearch cluster through the service.

In order to improve stability and reliability it would be really good to remove this configuration and let the operator dynamically change the configuration (Elasticsearch, fluentd, Kibana) based on the number of Elasticsearch instances in the cluster.