Updates for setting up the leaderboard

[yizhongw]

This PR contains several updates in order to set up the leaderboard here:

1. Add id to each instance in the Instances field. This is done by running the script in src/add_instance_id.py. Each id is composed of the task number and a string generated by uuid.uuid4().hex.
2. Update the official evaluation script in eval/automatic/evaluation.py. The new script uses GPT2 tokenizer for the xlingual track.
3. Add the script, Dockerfile, and configuration files for building the benchmark. These files are put in the eval/leaderboard folder.

[danyaljj]

Thank you, @yizhongw!