More flexible S3 object naming scheme

[mcandre]

kafka-connect-s3 currently includes information about Kafka partitions and offsets when formatting the S3 object name, correct?

My use case is more timestamp-oriented and less partition/offset-oriented. Could kafka-connect-s3 generalize the S3 object naming schema, such as allowing configuration to specify a class implementing an interface like String s3ObjectName(byte[] message)? That way, users could customize the S3 name to suit their backup system needs better.