Arm scale-to-zero cleanup before starting test compute

A one-pod Kubernetes qualification wrapper should arm EXIT and signal cleanup before the first scale-up mutation. Refuse a nonzero starting state, pin the exact pod UID, run once, then uninterruptibly scale to zero, wait for pod deletion, and verify both replica count and pod absence. Preserve the test failure status, but turn a cleanup failure into failure too. This prevents terminal or network interruption from silently leaving standby compute active; it does not make the test workload safe or replace admission fencing.

Applies when: A stopped Kubernetes workload is started temporarily for isolated migration or release qualification. Keywords: Kubernetes scale zero cleanup trap StatefulSet test runner teardown interruption