Clean `failed` pods

[kreynoldsf5]

For one reason or another, pods will fail (and the scheduler will recreate these). They are not being cleaned up and this is counting against storage quotas.

Fix this.

[kreynoldsf5]

We can't call the local kube-api from a pod with a serviceAccount. This makes sense -- the api endpoint we need to hit is really the vk8s endpoint.

We'll need to use a kubeconfig passed into a job pod.

[kreynoldsf5]

Let's do this from the spec itself.

For Jobs we can use ttlSecondsAfterFinished. We'll need to determine the correct way to handle for deployments.

[kreynoldsf5]

Using the following in the CronJon instead:

failedJobsHistoryLimit successfulJobsHistoryLimit