Bulk delete of all flows leaves flows dangling in the network

When a large number of flows is deleted from the config space in a single operation (e.g. a REST call with xpath 'restconf/config/opendaylight-inventory:nodes', which means delete all node configurations), the delete process stalls before all flows are deleted. Consequently, some flows are left in the network.

To reproduce:
1. Start the controller.

2. Start mininet with 3 nodes (OF1.0 or OF1.3 - does not matter, the issue
occurs with both)

3. Drive 32000 flows into mininet using the attached python script:
> python flow_config_perf.py --nflows 200 --nthreads=10 --ncycles 16 --no-delete

4. Validate in mininet that flows have been programmed into the network

5. Perform the bulk delete operation using the attached python script:
> python config_cleanup.py

6. Validate in mininet that flows have NOT been deleted (which would be
the expected outcome)