Tutorial on integration with fairseq-trained models

[neubig]

Fairseq is a widely used toolkit, so it'd be cool to have an example of how we can take the output of a fairseq run and upload it to ExplainaBoard through the client.