Check sqs queue length before checking job status

[vishalbollu]

Reverting order of operations to check queue length before job status because pods don't automatically terminate if one of the containers terminate. For a TensorFlow batch job, a pod has the batch container and the TF serving container. Even if the batch container exits, the TF serving container still remains running therefore the pod never exits. Since the pod never exists, job.active will be non-zero and the cortex job will remain in the running state.

Revert to checking the length of the queue for now until a better workaround is found.

---

checklist:

• run make test and make lint
• test manually (i.e. build/push all images, restart operator, and re-deploy APIs)