Database not initialised

[Robert Gildein]

After removing the leader unit, all other units went into a blocked state
with the message "Database not initialized".

```bash
juju status
Model Controller Cloud/Region Version SLA Timestamp
bug1864016 rgildein2-serverstack serverstack/serverstack 2.9.17 unsupported 15:38:47+01:00

App Version Status Scale Charm Store Channel Rev OS Message
keystone 9.3.0 waiting 2 keystone charmhub stable 538 ubuntu Database not initialised
openstack-dashboard 9.1.2 waiting 1 openstack-dashboard charmhub stable 514 ubuntu Incomplete relations: identity
percona-cluster 5.6.37 active 1 percona-cluster charmstore stable 302 ubuntu Unit is ready

Unit Workload Agent Machine Public address Ports Message
keystone/1* blocked idle 1 10.245.164.43 5000/tcp Database not initialised
keystone/2 blocked idle 2 10.245.163.171 5000/tcp Database not initialised
openstack-dashboard/0* waiting idle 4 10.245.164.13 80/tcp,443/tcp Incomplete relations: identity
percona-cluster/0* active idle 3 10.245.164.162 3306/tcp Unit is ready

Machine State DNS Inst id Series AZ Message
1 started 10.245.164.43 e63156f4-eeb9-4ccb-b600-abfba57d5821 xenial nova ACTIVE
2 started 10.245.163.171 227bd826-c210-4f80-ac30-faf5a6f8cc39 xenial nova ACTIVE
3 started 10.245.164.162 e3d6637a-01cc-4493-bf02-d292c890968f xenial nova ACTIVE
4 started 10.245.164.13 033dde99-5aa1-46d3-a7a1-abefe07775de xenial nova ACTIVE
```

I did the following [1].

---
[1]: https://pastebin.ubuntu.com/p/bvR56RhNZc/

[Robert Gildein (rgildein)]

The fix for this is to delete the `keystone:shared-db percona-cluster:shared-db`
relation and re-create it.