Scale Hyper-V Neutron Agent

A typical medium sized hybrid cloud deployment consists of more than
50 Hyper-V compute nodes along with computes like KVM or ESX.
The rate at which VMs are spawned/updated under such a deployment is
around 25 operations/minute. And these operations consists of spawning,
updation and deletion of the VM and its properties (like security group rules).
At this rate the possibility of concurrent spawn or update operations on a
given compute is High. What is typically observed is a spawn rate of ~2 VM(s)/minute.
Since WMI is not that performant, a VM port binding in Hyper-V neutron agent takes
10x amount of time when compared to KVM IPtable.
The situation worsens when the number of SG Rules to apply increases for a given port
(with the number of SG members), and there are many ports in queue to treat.
Under such a scenario neutron agent running on Hyper-v compute fails to complete binding
security rules to the VM port in given time,and VM remains in-accessible on the allocated
IP address.
This blueprint addresses the neutron port binding failure on the agent by
introducing port binding concurrency.