Ghislain Vaillant

See discussion in #413

This implementation passes the tests, but there are issues with type alignments between the nipype tasks and the BIDS reader task. I am investigating.

Looks like I found the issue. Testing on T1 linear works as expected.

I wonder whether we can do better than converting the query results to a tuple or single value and listing the keys to the output spec. Best would be to...

> I think it makes sense to add the functionality to filter on subject/session in this PR. That's gonna take longer then. Is that ok?

If I remove the sessions listed in the error message, then the pipeline launches successfully. I think the UX can be improved by still logging that some sessions may be...

I'd choose this over #699 considering ClinicaDL is already hosted on RTD.

> I don't know BrainStat well so I don't have a strong feeling about it being the right tool for the job or not. They are explicitly targeting themselves as...

We'd also need an additional `desc-preproc` entity as advised [here](https://bids-specification.readthedocs.io/en/stable/05-derivatives/02-common-data-types.html#examples) to distinguish the derived DWI filename from its original.

I have added `brainstat` to the project and refreshed the lock file. It brings quite a bit of transitive dependencies as a result. That's something to keep in mind. It...