extend the map function so it can wrap around long text that does not fit in the context window

### Feature request

I understand `dataset` provides a [`map`](https://github.com/huggingface/datasets/blob/main/src/datasets/arrow_dataset.py#L2849) function. This function in turn takes in a callable that is used to tokenize the text on which a model is trained. Frequently this text will not fit within a models's context window. In this case it would be useful to wrap around the text into multiple rows with each row fitting the model's context window. I tried to do it using this code as example which in turn I have borrowed from [here](https://stackoverflow.com/a/76343993/147530):

```
data = data.map(lambda samples: tokenizer(samples["text"], max_length=tokenizer.model_max_length, truncation=True, stride=4, return_overflowing_tokens=True), batched=True)
```

but running the code gives me this error:

```
File "/llm/fine-tune.py", line 117, in
data = data.map(lambda samples: tokenizer(samples["text"], max_length=tokenizer.model_max_length, truncation=True, stride=4, return_overflowing_tokens=True), batched=True)
File "/llm/.env/lib/python3.9/site-packages/datasets/arrow_dataset.py", line 580, in wrapper
out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs)
File "/llm/.env/lib/python3.9/site-packages/datasets/arrow_dataset.py", line 545, in wrapper
out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs)
File "/llm/.env/lib/python3.9/site-packages/datasets/arrow_dataset.py", line 3087, in map
for rank, done, content in Dataset._map_single(**dataset_kwargs):
File "/llm/.env/lib/python3.9/site-packages/datasets/arrow_dataset.py", line 3480, in _map_single
writer.write_batch(batch)
File "/llm/.env/lib/python3.9/site-packages/datasets/arrow_writer.py", line 556, in write_batch
pa_table = pa.Table.from_arrays(arrays, schema=schema)
File "pyarrow/table.pxi", line 3798, in pyarrow.lib.Table.from_arrays
File "pyarrow/table.pxi", line 2962, in pyarrow.lib.Table.validate
File "pyarrow/error.pxi", line 100, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: Column 1 named input_ids expected length 394 but got length 447
```

The lambda function I have provided is correctly chopping up long text so it wraps around (and because of this 394 samples become 447 after wrap around) but the dataset `map` function does not like it.

### Motivation

please see above

### Your contribution

I'm afraid I don't have much knowledge to help