RabbitMQ message sizes may get very large and cause problems

[PasiSa]

When Radar matchlib task has processed the submissions for an exercise, it sends the results back to Radar service in a single RabbitMQ message. If the number of submissions is very large, such as in some exercises of the Y1 course (> 2800 submissions), this causes a large RabbitMQ message between the Celery tasks, that exceeds the default 128 MB maximum message size. In addition, it causes very large memory consumption on RabbitMQ. For now, this problem can be prevented by increasing the max_message_size setting in rabbitmq.conf, but in longer run a better solution in Radar is needed, to distribute the result in smaller messages, and in general to avoid large messages between tasks.