Refactor wait_for_cpd_pod.sh to poll job status instead of pod

[ocofaigh]

1. The current wait_for_cpd_pod.sh script is polling the status of the first pod in the namespace. It would be safer to poll the status of the job instead just incase other pods some day run in the same namespace.

2. The namespace is hard coded in the script but really we should be passing it to the script based on what is passed to the helm chart.

3. The logic only checks if pod is in a Failed or Succeeded state. Otherwise it will run in an infinite loop. We should probably add more advanced logic here in case of other errors (such as image pull).