Fix designate upgrades when internal RPC version changes

When upgrading from 2023.1 to 2023.2 in a HA environment,
the Designate pool update command failed as it ran using
an upgraded client, but made RPC calls against instances
which had not been upgraded yet.

This commit moves the pool update to the last play host
in order to ensure that in an upgrade situation, all
virtualenvs have been upgraded first to ensure no client/
server version conflicts.

Change-Id: I93af581696de0913710592f9033545b87547b78a

tasks/main.yml

@@ -198,7 +198,7 @@
     group: "{{ designate_system_group_name }}"
   when:
     - designate_pools_yaml is defined
-    - _designate_is_first_play_host
+    - _designate_is_last_play_host
   notify:
     - Perform Designate pools update
   tags:

vars/main.yml

@@ -19,6 +19,12 @@ _designate_is_first_play_host: >-
       inventory_hostname == (groups[designate_services['designate-api']['group']] | intersect(ansible_play_hosts)) | first) | bool
   }}
 
+_designate_is_last_play_host: >-
+  {{
+    (designate_services['designate-api']['group'] in group_names and
+      inventory_hostname == (groups[designate_services['designate-api']['group']] | intersect(ansible_play_hosts)) | last) | bool
+  }}
+
 _designate_oslomsg_rpc_vhost_conf: >-
   {{
     (designate_oslomsg_rpc_vhost is string) | ternary(