The performance of quota usage sync in Neutron driver should be enchanced

Currently the quota usage is sync-ed through list of all resources of these project and then sum. This handling's performance is not good, especially if multiple quota-usage task executed concurrently, will heavily impact the performance of Neutron for many queries in parallel

                networks = self.neutron.list_networks(**opts)['networks']
                subnets = self.neutron.list_subnets(**opts)['subnets']
                ports = self.neutron.list_ports(**opts)['ports']
                routers = self.neutron.list_routers(**opts)['routers']
                floatingips = self.neutron.list_floatingips(
                    **opts)['floatingips']

                usages['network'] = len(networks)
                usages['subnet'] = len(subnets)
                usages['port'] = len(ports)
                usages['router'] = len(routers)
                usages['floatingip'] = len(floatingips)

After Neutron client (or OpenStackClient) support limit calling to get the usage directly, the code here should be improved.