Allow configuration of failed reconcile handling

[Jay-Madden]

As documented in #224 the ability to specify say a backoff pattern for errors when the reconciler is mounted would prevent the operator from incessantly spamming logs and the kubeapi attempting to reconcile a CRD that will never work.

Leading to logs that look like this

This is work that we would be willing to contribute as a community contribution if the idea is accepted as a desirable feature.