Kafka Consumer Event Loop

[nomisRev]

To be able to properly stream Kafka you need to have an event loop that properly facilitates streaming Kafka.

There are several reasons why this is needed:

• Commits can only occur during poll events
• You need to poll continuously, or pause partitions to guarantee proper back-pressure without expensive rebalancing/repartitioning
• Optimising committing offsets to Kafka with strong guarantees in face of stream termination

Also see https://tuleism.github.io/blog/2021/parallel-backpressured-kafka-consumer/.

This PR adds a custom event loop for facilitating this based on reactor-kafka.
Some work still needs to be done to support EXACTLY_ONE and AT_MOST_ONCE delivery.

Given this work makes everything in the Consumer.kt file obsolete I decided to move it to a .receiver package, following similar naming as reactor-kafka but this is prone to change/improvement towards 1.0.

I tried to add as much documentation inside of the code as possible. All feedback, and suggestions, code reviews or questions are welcome! 🙏