Validate websocket connections when we scale the pods on the websockets deployment

In production#3420 (closed) we saw three pods come online, where there was a brief spike of errors at the load balancer. The best theory we have right now is that it coincided to a spike of traffic, which resulted in 3 new pods coming online. During the time when the pods were coming online it looks like we saw traffic routed to them before they were ready.

We need to experiment some more to see what exactly is going on here, are we passing the readiness before we are ready? Is it something else?