Q -> R: ceph mgr down after upgrade due to start-limit-hit

[Peter Sabaini]

**Description:**
While upgrading `ceph-mon` from Quincy to Reef, I encountered an issue where `ceph-mgr` restarts too quickly. This leads to hitting the start limit for `systemd`.

This does not appear to be consistent though, on two consecutive runs I've first seen 3 of 3 mgrs down, on the next run only 1 of 3 was down.

**Reproduction Steps**
1. Deploy quincy cloud
2. Run `juju config ceph-mon source=cloud:jammy-bobcat`.

**Error Message:**
When checking the status using `sudo systemctl status <email address hidden>`, this error was shown:

```shell
ubuntu@juju-bc9f56-zaza-5ec88f2270ac-7:~$ sudo systemctl status <email address hidden>
× <email address hidden> - Ceph cluster manager daemon
...
Oct 05 09:02:11 juju-bc9f56-zaza-5ec88f2270ac-7 systemd[1]: <email address hidden>: Start request repeated too quickly.
Oct 05 09:02:11 juju-bc9f56-zaza-5ec88f2270ac-7 systemd[1]: <email address hidden>: Failed with result 'start-limit-hit'.
Oct 05 09:02:11 juju-bc9f56-zaza-5ec88f2270ac-7 systemd[1]: Failed to start Ceph cluster manager daemon.

```

**Workaround:**
Reloading `systemd` seems to solve this, as the service starts correctly after running `sudo systemctl daemon-reload`.

**Additional Information**
The `charm` version was latest/edge at git revision 55beb25.