Description of problem:
In the compute templates, we assign the bridge_name parameter to the value of NeutronPhysicalBridge. This allows the --neutron-physical-bridge parameter on the command line to influence the bridge name in the NIC templates. Unfortunately, in the controller templates we hard-code the value to "br-ex", so if you use the --neutron-physical-bridge parameter your compute nodes will not have the same bridge name as your controller nodes, and VLAN mode networking won't work.

Version-Release number of selected component (if applicable):
7 GA

How reproducible:
100%

Steps to Reproduce:
1. Use --neutron-physical-bridge command-line parameter
2. This sets NeutronPhysicalBridge, which sets bridge_name in compute tempates
3. Controller templates hard-code br-ex, so the command-line parameter fails to work.

Actual results:
Controllers *always* end up with br-ex as the name of the bridge where the VLANs terminate.

Expected results:
This should be influenced by a command-line parameter --neutron-physical-bridge

Additional info:
This was discovered while triaging a separate bug. While both of these bugs are in place, there is no way to make VLAN mode networking work (unless you replace the name of the bond "br-bond" in the compute.yaml with "br-ex" or "{get_input: bridge_name}".

controller-puppet.yaml L585:
        bridge_name: br-ex