Elasticsearch scaling should take redundancyPolicy into consideration

1. Proposed title of this feature request
With Cluster Logging Operator (CLO) it's possible to scale elasticsearch up and down depending on customer needs. Problem is, depending on redundancyPolicy applied the customer could loose data when scaling down as the redundancyPolicy might hold data just on the elasticsearch nodes that are being removed.

This is why redundancyPolicy should be taken into consideration when scaling down to prevent data loss from happening.

2. What is the nature and description of the request?
This is a follow-up from RHBZ #1879150 where the scaling down functionality was improved. Based on this work it was found that it would scale down no matter what redundancyPolicy is set and therefore leaving the elasticsearch cluster in a bad state.

3. Why does the customer need this? (List the business requirements here)
As ŵorkloads can dynamically move between OpenShift - Cluster and also change, it may be necessary to scale elasticsearch cluster up but in some cases also down. Even though documentation should cover the problems around scaling down, we should try to prevent data loss from happening with technical implementation as documentation may not always be considered when scaling.

4. List any affected packages or components.
Cluster Logging Operator and Elasticsearch Operator