Description of problem:

The imagePullPolicy of the adp-controller-manager and velero pods are "Always". This is problematic in edge scenarios where there may be limited network bandwidth to the registry resulting is slow recovery time following a pod restart.  

The general openshift recommendation is to use policy{} IfNotPresent. This jira request this update be done to OADP.

For OADP 1.3, change imagePullPolicy based on container image. If it contains a digest, use IfNotPresent; otherwise, Always. User can not change that value manually, reconcile loop will just update value.