number of "Core flows" drastically increases when restarting a node running fabric

Seems like maybe this is two issues, 
1. # of flows shouldn't increase when restarting a node
2. Some of these flows probabaly shouldn't be under the "core" appId

Core Flows from one switch are here:
https://docs.google.com/document/d/1lzt7tZsFdttU0587C36Dw3pGBknz0NDSDz4Q653Mj1c/edit?usp=sharing

Also including onos-diagnostics tarball.

Environment:

On commit:

• 5f99ef2 (HEAD -> onos-1.12) Routing/bridging rules on the same leaf pair should always be programmed by the same ONOS instance
• 03cab54 (origin/onos-1.12) Fixes for maven build

7 node cluster with this topo:

sudo -E python ~/mininet/custom/comcast_fabric.py --onos-ip=10.192.19.211,10.192.19.212,10.192.19.213,10.192.19.214,10.192.19.215,10.192.19.216,10.192.19.217  --dhcp=1 --routers=1 --ipv6=0 --ipv4=1