Rely on kafka partitioning to aid parallel execution

Since 1.3.3 eventutilities is aware of message_key_fields its serializer and encodes update row objects as records with the specified fields as JSON key. Once we partition the update topic, we get parallelisation support by the flink-kafka-source out of the box, as soon as we launch the app with replicas (= task slots in helm context) > 0.

Bug: T354064 Change-Id: I045b7837d49746a7d485a4360728408822b65c5a