Fail job immediately on CreateContainerConfigError in k8s executor (closes #39361)

What does this MR do?

Fails the Kubernetes executor job immediately when a container enters CreateContainerConfigError state.

Why was this MR needed?

The pod watcher's CheckTerminalContainerErrors handled InvalidImageName, ErrImagePull, ImagePullBackOff, and CrashLoopBackOff, but not CreateContainerConfigError. Jobs referencing a missing ConfigMap, Secret, or PVC would hang indefinitely instead of failing fast.

What's the best way to test this MR?

Create a Kubernetes executor job with a pod referencing a non-existent ConfigMap. The job should fail with a clear configuration error rather than hanging until timeout.

What are the relevant issue numbers?

Closes #39361 (closed)