Add null prompt for gsarti/flores_101

[cjlovering]

gsarti/flores_101 is a dataset with 102 languages used for language modeling.

• Added the 102 language subsets
• The prompt is empty -- just the sentence itself -- as this dataset will be used for language modeling (LM).
• The metric is set to Other; downstream applications (eval harness) will select the LM metric.
• It was necessary to add gsarti to the user list because the dataset is listed under that user in hugging face datasets.

[cjlovering]

Given that this is a null prompt and we're currently using it as as a language modeling dataset, we can skip this for now. Using promptsource isn't necessary.

[stephenbach]

Can this PR be closed if we're not prompting it?

[rbawden]

Hi there! For information, I have just prompted this dataset (automatically for MT). Linking the PR here: #779