Improve error message in collectd plugins when they cannot authenticate to Keystone

[Simon Pasquier]

When the collectd plugins can't authenticate to Keystone, the collectd logs are filled with messages like:

[2015-11-26 10:35:02] Cannot get a valid token from http://openstack.example.com:5000/v2.0
[2015-11-26 10:35:02] Service 'keystone' not found in catalog
[2015-11-26 10:35:02] Could not find Keystone tenants
[2015-11-26 10:35:02] Cannot get a valid token from http://openstack.example.com:5000/v2.0
[2015-11-26 10:35:02] Service 'nova' not found in catalog
[2015-11-26 10:35:02] Could not get hypervisor statistics
[2015-11-26 10:35:02] Cannot get a valid token from http://openstack.example.com:5000/v2.0
[2015-11-26 10:35:02] Service 'cinder' not found in catalog
[2015-11-26 10:35:02] Could not find cinder volumes
(snip)
[2015-11-26 10:36:02] Cannot get a valid token from http://openstack.example.com:5000/v2.0
[2015-11-26 10:36:02] Service 'keystone' not found in catalog
[2015-11-26 10:36:02] Could not find Keystone tenants
...

I suggest several improvements to ease troubleshooting:
- Improve the "Cannot get a valid token" message:
  Cannot get a valid token from http://openstack.example.com:5000/v2.0. Check that the OpenStack credentials (tenant=services, user=nova) are valid.
- Do not log the "Could not find ..." message or log it at a different level
- Reduce the message frequency, the same messages are logged every minute.