Investigate why we sometimes have multiple pods per gameserver

[jkowalski]

During large cluster turnup I ran into this:

$ kubectl get gs
NAME                     STATE          ADDRESS   PORT   NODE   AGE
simple-udp-sv722-gj985   RequestReady                           27m
simple-udp-sv722-m7rk2   RequestReady                           27m

$ kubectl get pod
NAME                           READY   STATUS    RESTARTS   AGE
simple-udp-sv722-gj985-csgr7   2/2     Running   0          27m
simple-udp-sv722-gj985-rw6p7   2/2     Running   0          27m
simple-udp-sv722-m7rk2-g8rn8   2/2     Running   0          27m
simple-udp-sv722-m7rk2-q44mm   2/2     Running   0          27m

In a large turnup (10K GS) this reproes quite often - typically one or two of the GS get stuck in this state.