Allow converting entire NV collections as Datasets

[adelavega]

https://github.com/neurostuff/NiMARE/blob/c3f26675709048ae9558be6223f5e0f50741c0ca/nimare/io.py#L451

Here you have to specify which contrasts to convert from collections.

Would it make sense to also allow "all". As in, if contrasts=None, just ingest all candidate images and tag them with the name in NeuroVault?

Also, I'm a little worried the path name for images is not necessarily unique (it's just the contrast names concatenated). I'm not sure if that matters much since file names should be but just saying.

[tsalo]

I think @jdkent is the best person to make the final decision on this, but that sounds like a good idea to me.

Also, I'm a little worried the path name for images is not necessarily unique (it's just the contrast names concatenated). I'm not sure if that matters much since file names should be but just saying.

I'm thinking that we should update all of our filenames to be more consistently BIDS-like, so we can work on ensuring that filenames are unique at the same time. I could be wrong, but it seems like the downloaded files are unique, since they have the collection and image IDs in the filenames, so the files generated by ImageTransformer are the problematic ones. Does that sound right?

[adelavega]

The file names are unique, but they are organized into folders such as "contrast1_contrast2", which I are not guaranteed to be unique. But if there's overlap the individual files should be unique hence why I'm not sure it matters much.