I have a long run spark structure streaming job to read data for a Kafka topic. I am using s3 as the kafka checkpoint storage. It works well. However, I saw a lot files were cre