Limit volume size of submissions

[rrchai]

• fixes #23 The idea here is to limit the size of submission. This pr is to create a volume (output_volume) first with limited size and copy over the prediction files to the working dir of run docker. If the volume usage is beyond the size limit, the model will break - error should be in the model log file.

Also, before the model failed, the output volume is not cleaned up until successful submissions. Now, the volume can be easily removed after failure.

> docker inspect  9729442

        "Mounts": [
            {
                "Type": "bind",
                "Source": "/home/ec2-user/challenge-data/downsampled/scrna-synthetic",
                "Destination": "/input",
                "Mode": "ro",
                "RW": false,
                "Propagation": "rprivate"
            },
            {
                "Type": "volume",
                "Name": "9729442",
                "Source": "/var/lib/docker/volumes/args.submissionid/_data",
                "Destination": "/output",
                "Driver": "local",
                "Mode": "rw",
                "RW": true,
                "Propagation": ""
            }