debug should set deployment strategy=recreate

Consider configuring [deployments to use a `recreate` strategy](https://kubernetes.io/docs/concepts/workloads/controllers/deployment/#recreate-deployment), similar to how we reduce the replica counts to 1. This causes the deployment controller to terminate old pods before creating new pods. This should avoid any possibility of having stale pods running old code being available in the system.

We should also set [`revisionHistoryLimit: 0`](https://kubernetes.io/docs/concepts/workloads/controllers/deployment/#revision-history-limit) to reduce the memory pressure required to retain previous deployments. People aren't going to do rollbacks from `debug`.