Correct way to consume multiple files

What is the correct way to consume multiple file?

I've implemented a `RandomAccessDataSource` which opens and pre-processes each file.

To me the benefit of grain is that I don't have to manually implement batching or sequences (for RNN), and grain knows how to shape the arrays accordingly. If the solution to consuming multiple files is that I have one `RandomAccessDataSource` per file, wrap those in `DataLoader`s, and then for loop over the `DataLoader`s/files, then I lose the batching benefit - I'm effectively having to implement batching.

Is there a way of telling a `DataLoader` "your data is these N `RandomAccessDataSource`"?