TODO add example of HPA based on consumer lag

[solsson]

Horizontal Pod Autoscaler being too long for an issue title :)

Great idea from @StevenACoffman in #147 (comment) as it nicely would combine features from Kubernetes and Kafka, and make real use of the ubiquitous consumer lag monitoring (instead of just exporting a metric).

I think I've seen hints that HPA can be based on custom metrics from Prometheus. Due to #125 we do have such metrics.