Kubernetes job watching does not work when erroring

[Munsio]

When we are running multiple models within the kubernetes runner we are only watching for the completed type on the job. When an model is not available anymore the job is going into an error state which we also need to take into account.