Hello Team, We're introducing Apache Druid into our organization. However, we're unable to use the default extension "_*druid-kafka-indexing-service*_" for real-time data intake. Could someone advise us on creating a new extension for real-time data intake? Or suggest modifying the existing "druid-kafka-indexing-service" extension to use the "_*KafkaConsumer.subscribe()*_" API instead of "_*KafkaConsumer.assign()*_"? If there's another extension for real-time data intake that uses the "KafkaConsumer.subscribe()" API, please recommend it as well. Thank you.

You really want to use assign instead of subscribe. It stores the offsets in the data itself as it is ingested, giving many guarantees that it isn't missing or duplicating data. A subscribe based ingestion would be a strict downgrade.