e2e test 'restart-counted' is flaky

[sharnoff]

Hit this a few times over the past few weeks. Finally got around to opening an issue for it after hitting it again. A couple examples:

• https://github.com/neondatabase/autoscaling/actions/runs/9372090123/job/25802975106#step:13:858
• https://github.com/neondatabase/autoscaling/actions/runs/9373078189/job/25807554041#step:13:866

Seems like in general the issue is that the VM has Phase=Running and RestartCount=0 after the pod has been killed.

At first I thought it might be a race between when the controller updates the VM after restarting and when kuttl checks... but it seems like the issue is actually that it's still that way 30 seconds later?