lbaas update member status causing dead locks in db

There is a thread calling _update_member_status() every 5 seconds in loadbalancer v2 driver. With significant amount of active load balancers, and 3 redundant controller nodes, that causes heavy load on the database. After some point, this process stops fitting in 5 seconds, queries stack on each others and cause dead locks. Neutron ends up eating 100% CPU and stops functioning properly.