Allow "jsonl" data type signifier

### Feature request

`load_dataset` currently does not accept `jsonl` as type but only `json`.

### Motivation

I was working with one of the `run_translation` scripts and used my own datasets (`.jsonl`) as train_dataset. But the default code did not work because

```
FileNotFoundError: Couldn't find a dataset script at jsonl\jsonl.py or any data file in the same directory. Couldn't find 'jsonl' on the Hugging Face Hub either: FileNotFoundError: Dataset 'jsonl' doesn't exist on the Hub. If the repo is private or gated, make sure to log in with `huggingface-cli login`.
```

The reason is because the script has these lines to extract the data type by its extension. Therefore, the derived type is `jsonl` which is not recognized by datasets as the error above shows.

https://github.com/huggingface/transformers/blob/ade26bf9912f69e2110137443e4406d7dbe253e7/examples/pytorch/translation/run_translation.py#L342-L356

I suppose you could argue that this is the script's fault (in which case I'll do a PR over at `transformers`) but it makes sense to me to add `jsonl` as an alias to `json` in `datasets`.

### Your contribution

At the moment I cannot work on this. I think it can be as "easy" as having an alias for json, namely jsonl.