Add preprocessing for experiments without spin-echo fieldmaps

[mwaskom]

This will probably require a completely separate preprocessing workflow (but using many of the custom interfaces developed in #106).

Should this workflow be oriented towards non-SMS acquisitions and thus also not allow a separate reference EPI volume? Alternatively, this could be an option, but it could register everything to the first volume if it's not availible. But that gets complicated! Could just expect people to slice off their own first volume and give that if they don't have a separate reference image.