Pod restarts stop after raising memory, but rollout still fails

A team fixes the direct crash cause and still cannot complete rollout because the health timing contract was never updated.