Make resources (requests + limits) of the Operator Pods configurable persistently

Currently the resources available to the operator pods (e.g. apicast-operator-controller-manager, threescale-operator-controller-manager) are only configurable in the csv (cluster service version) yaml definition.

Configuring the values this way requires then to reconfigure them every time the Operator is upgraded because the CSV will not persist unchanged after the upgrade procedure, and therefore such values would be lost.

The request is to make the resources configurable persistently so that the configuration can survive the Operator upgrades.