[KafkaScaler] Scale beyond the number of partitions

[lionelvillard]

Proposal

Add an option to disable the check that prevent KafkaScaler to scale beyond the number of partitions

Use-Case

When KEDA is used to scale CRDs, the number of replicas does not necessarily match the number of Kafka consumers. For instance, KEDA can we used to increase/decrease the throughput (msgs/sec). In this usecase it makes sense to have more replicas than partitions.

[zroubalik]

Fine by me!
@tomkerkhove wdyt?

[tomkerkhove]

Fine by me.