What would you like to be added:
Currently, CCM uses exponential backoff when failing to process a service change. It would be great if this behavior—not only the minimum and maximum delay values—could be adjusted if needed.

Why is this needed:
It is possible to change the minimum and maximum delay values but in some cases that's not enough. Changing those values affects all kinds of service changes, and it would be great if it were possible to adjust that behavior in specific circumstances.

Take for example a new Load Balancer being created. We might want to reduce the max retry delay to just a few seconds long so it's ready within the cluster with minimal delay, but for other types of sync failures, keep the max retry delay set to 5 minutes as to not overload the API endpoint that is being polled.

/sig cloud-provider

/cc @timoreimann — Timo, could you please take a look and make sure this is in line with what we discussed?