'admin' client name is hardcoded

[Przemyslaw Hausman]

'admin' client name is hardcoded here (`svc = 'admin'`):
https://opendev.org/openstack/charms.ceph/src/branch/master/charms_ceph/broker.py#L852

This creates a problem when a ceph charm (e.g. ceph-radosgw) is related with ceph-proxy (instead of ceph-mon) and ceph-proxy charm is configured with `admin-user` different than 'client.admin'.

As a result, ceph commands run on ceph-proxy unit fail, because ceph-proxy is using different client than 'admin', e.g. 'my-ceph-proxy-admin'. See example error message from ceph-proxy:

unit-ceph-proxy-1: 12:27:47 ERROR unit.ceph-proxy/1.juju-log radosgw:9: Command '['ceph', '--id', 'admin', 'config-key', 'put', 'cephx.groups.objects', '{"pools": ["demo-15658.rgw.buckets.data"], "services": []}']' returned non-zero exit status 1.

Steps to reproduce:

```
# Deploy ceph-radosgw and ceph-proxy
juju deploy ceph-radosgw \
  --config source='cloud:focal-yoga' --channel quincy/edge
juju deploy ceph-proxy \
  --config source='cloud:focal-yoga' --channel quincy/edge

# Configure ceph-proxy with custom admin-user
juju config ceph-proxy \
  fsid=4d6ef0f6-dc23-11ec-a706-294ff47d2a61 \
  admin-user=client.my-ceph-proxy-admin \
  admin-key='AQD/J49ifTrqJBAAe5vxsTxWcedzKrRZPnMyhw==' \
  monitor-hosts="172.27.80.175:6789"

# Relate ceph-radosgw and ceph-proxy
juju add-relation ceph-proxy:radosgw ceph-radosgw:mon
```

ceph-radosgw unit will be stuck in blocked workload state with message "Services not running that should be: <email address hidden>". And in the debug-log for ceph-proxy unit you'll see the error message mentioned earlier.