[FIX] don't resample subject-space ROIs unless user provides something

[36000]

Currently I think subject-space ROIs are resampled to MNI, which is a bug.
It used to say:
"
If None, the template will be overriden when passed to
an API class."
But I see no evidence that I actually implemented this.

[arokem]

Do we need to add tests to keep this from breaking?

[36000]

I think we should. do we have some subject-space ROIs for this subject we could use?

[arokem]

There's a functionally-defined LV1 for the Stanford HARDI dataset here: https://searchworks.stanford.edu/view/ng782rw8378

This notebook shows how it could be downloaded: https://github.com/arokem/visual-white-matter/blob/master/download-data.ipynb

[36000]

What does LV1 mean? lower primary visual cortex? Than should the other ROI between the LGN for the bundle?

[arokem]

It's left visual cortex. Maybe we can find another ROI in the aparc reduced file that is also in that same collection?

[36000]

@arokem I brought in your changes from the PR in my repository. This is now ready for review

[36000]

@arokem this is ready for review / merge

[arokem]

Looks good. At some point we should also make an example of the subject-space ROI functionality for endpoints, since this is a pretty cool use-case.