Praefect subcommand to initiate an immediate failover to a specific shard

As things stand right now, Praefect failover only happens if the primary is unhealthy for 10 seconds, and we don't have control over which node gets to be the next primary.

Having the option to initiate a failover at will (without waiting for a timeout) and to a specific shard is beneficial for maintenance purposes, such rolling kernel updates to shards.