Rotating oblique volumes and bvecs

[josephmje]

I've been doing more tractography lately and was getting really bad tract segmentation on a particular group of scans acquired on a Siemens Prisma. After talking to Yogeth Rathi's group and @mattcieslak , both noted that it could be an alignment issue.

Matt has also noticed something similar in the HBN dataset and has implemented a correction in qsiprep. I implemented a fix on our lab's data using a pipeline from Penn BBL.

However, going forward, I'm unsure how dmriprep handles rotating the bvecs. Is there a place already in the code that I should be modifying?
@arokem @dPys @oesteban

This might also fix the error in #113

[dPys]

@josephmje
This is probably what you're looking for: https://github.com/nipreps/dmriprep/blob/master/dmriprep/utils/vectors.py#L479-L485