Collector can not persistence compute node CPU metrics when set ‘compute_monitors = ComputeDriverCPUMonitor’ in nova

Reproduce step:
1.Set ‘compute_monitors = ComputeDriverCPUMonitor’ in nova.conf
2.Start the ceilometer-collector,ceilometer-agent-notification and nova-compute
Then you can see the following error trace in ceilometer-collector console:
http://paste.openstack.org/show/73874/

The root cause of that issue is the counter_type is the 'tuple', In actually the type of the counter_type should be String, so it can not persistence to DB success, please refer to the SQL, the counter_type value is [u'cumulative'] that is a 'tuple'

INSERT INTO meter (counter_name, user_id, project_id, resource_id, resource_metadata, counter_type, counter_unit, counter_volume, timestamp, message_signature, message_id) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)('compute.node.cpu.idle.time', None, None, 'localhost.localdomain_localhost.localdomain', '{"name": "cpu.idle.time", "timestamp": "2014-03-20T04:09:59.462867", "value": 25406680000000, "source": "libvirt.LibvirtDriver", "host": "compute.localhost.localdomain", "event_type": "compute.metrics.update"}', [u'cumulative'], 'ns', 25406680000000, Decimal('1395288599.462867'), '425bec815573bf6b66ce0103aa5a61c879a2902397b0733428103d5be5a811c4', '80d34bfe-afe5-11e3-9564-0c82685348ee')