Thanks, I actually figured it out...
The Edge and Management cluster aren't transport node servers and I didn't have any VMs on the workload cluster that have hosts configured as transport nodes so when I took an Edge node down everything came down.
I'm guessing you have to have so many transports nodes up with a tunnel to make failover work.
I created VMs in the workload cluster that brought the tunnels up and when I put an edge node in maintenance everything kept humming along.
I guess it's looking for a heatbeat from other transport nodes to make sure it's worth failing over otherwise it just stays down.