I, too, would like my primary shards to be balanced across the cluster.
I often reduce the number of replicas on older indices to optimize disk space.
When the cluster is reasonably idle, shards will recover from just-maked-for-deletion copies.
But quite often this means hours of indices being relocated to nodes that just deleted them.
Search engine landed me here: rebalance-primary-shards/14470
Best explanation of problems is here: how-to-rebalance-primary-shards-on-elastic-cluster/176060/4
The requirement for rebalancing was also asked about in primary-shard-rebalancing/6173 and cluster-reroute-automatically-reroute-and-rebalance-all-primary-shards/161757
Is this the proper way to inject solutions into older topics?