Support >1 PWC dataset per dataset card

**Is your feature request related to a problem? Please describe.**
Some datasets cover more than one dataset on PapersWithCode. For example, the OffensEval 2020 challenge involved five languages, and there's one dataset to cover all five datasets, [`strombergnlp/offenseval_2020`](https://huggingface.co/datasets/strombergnlp/offenseval_2020). However, the yaml `paperswithcode_id:` dataset card entry only supports one value; when multiple are added, the PWC link disappears from the dataset page.

Because the link from a PapersWithCode dataset to a Hugging Face Hub entry can't be entered manually and seems to be scraped, this means end users don't have a way of getting a dataset reader link to appear on all the PWC datasets supported by one HF Hub Dataset reader.

It's not super unusual to have papers introduce multiple parallel variants of a dataset and would be handy to reflect this, so e.g. dataset maintainers can DRY, and so dataset users can keep what they're doing simple.

**Describe the solution you'd like**
I'd like `paperswithcode_id:` to support lists and be able to connect with multiple PWC datasets.

**Describe alternatives you've considered**
De-normalising the datasets on HF Hub to create multiple readers for each variation on a task, i.e. instead of a single `offenseval_2020`, having `offenseval_2020_ar`, `offenseval_2020_da`, `offenseval_2020_gr`, ...

**Additional context**
Hope that's enough

**Priority**
Low