Hive: * Remember last local actor id even when killing (but not yet deleting) node * Rename current restart statistics to disconnect statistics, keep using them for any "bad node" logic * Add genuine restart statistics - ones that come with actor id change * Add new field(s?) to node stats message Healthcheck: * Simply use the new fields