Investigate many `updated` messages in reconcile loop

Summary

In the MRs listed below, we observed that upon deploying the Operator and a GitLab CR, the reconcile loop would continuously print that it had updated various objects long after the GitLab CR reported Running and all the objects were present and ready.

• !163 (comment 599505513)
• !169 (comment 604066456)

This means the logs are flooded with messages, making it hard to notice any other important information there. It doesn't appear that any objects like pods are actually being updated (for example, pods aren't being repeatedly terminated and restarted).

On a related note, it's my understanding that the controller should wait ~5 seconds before trying the reconcile loop again if it's waiting for resources to finish coming up. In general, it would be good to research and document the behavior of the Operator in terms of what triggers a reconcile loop and how long it waits between loops. Perhaps predicates (#201 (closed)) would be helpful here as well.

Acceptance criteria

• Events and/or situations that cause a reconcile loop are documented
• Reconcile loop does not repeat endlessly saying certain objects are updated #295 (closed)
• NGINX Ingress Controller's Service object does not repeatedly get new nodePort values