Remove role re-balancing

[carlcsaposs-canonical]

If there are an even number of cluster-manager-eligible nodes, OpenSearch automatically excludes one from the voting configuration

There should normally be an odd number of master-eligible nodes in a cluster. If there is an even number, Elasticsearch leaves one of them out of the voting configuration to ensure that it has an odd size.

https://www.elastic.co/guide/en/elasticsearch/reference/current/modules-discovery-voting.html#_even_numbers_of_master_eligible_nodes

Therefore, there is no need to ensure an odd number of cluster-manager-eligible nodes by re-balancing the roles ourselves

---

This solves the issue with the deterministic role re-balancing in #209 where:

• initial deployment
• even number of units
• leader unit is highest unit number
• (highest unit number is not cluster manager eligible)
• (leader unit has to start first, but must be cluster manager eligible)

Alternative solutions to that issue:

• start highest unit as cluster manager eligible and restart it later to remove cluster-manager-eligible role (cons: extra restart, at what point do you restart unit—requires special case for initial start, could affect HA during initial start)
• replace deterministic role re-balancing with non-deterministic role re-balancing (cons: upgrade cannot be coordinated without use of peer databag which harms rollback resiliency—detailed in Re-balance roles deterministically instead of randomly #209 (comment) or without race for lock [which has other concerns & deviates from upgrade design used in other charms])

This also removes restarts (that were needed on a unit when its roles changed)

Context: https://chat.canonical.com/canonical/pl/756bhdey33ysjx3qdee3ktgoxo

[github-actions]

https://warthogs.atlassian.net/browse/DPE-4057

[carlcsaposs-canonical]

236ff8f in #211 contains initial fix for this issue

Complete fix (that removes unnecessary code) planned to be implemented by @Mehdi-Bendriss in a separate PR