Fix scaling when using ingress-addressed nodes

[HoustonPutman]

Fixes #682

This PR does two major things that fixes scaling issues with ingress-addressed node SolrClouds.

• Whenever a cluster is scaled down, the hostAliases that given to the Solr Pods are not updated. The hostAliases are only changed if a new host alias is added or an existing host alias is changed. There's no need to do a rolling restart to remove a host alias that will no longer be used.
• When a cluster is scaled up, the hostAliases must be updated to include the new pods. This requires a rolling restart, which we want to occur before trying to rebalance the cluster, since the new pods need to be addressable from the old pods (which the rolling restart will enable). Therefore its best to wait to scale until the rolling restart is complete.

[gerlowskija]

LGTM!