Add infos about datasets used for evaluation

We need to add in the README infos about the datasets used for evaluation (ie XSUM & OrangeSum)

To do in repo :

• Link to paper
• Link to repo (GitHub, HF, etc)

To do for publication:

• Consolidate tables
• Add annotations to the dataset (language, lengths of texts, words, diversity, etc)

We do not republish the dataset, only ref (and method on how we use the dataset for our task)