Workflow for multiecho data

[emdupre]

I see in #808 that you're encouraging the creation of a unique workflow instead of modifying the init_func_preproc_wf as I've been doing in #875. Do you have a preference for either creating unique workflows for specific cases (like functional-only or multi-echo scans) or modifying existing workflows in place?

[oesteban]

Well, I think your route is better. In #808 we took that path because it seemed easier for new contributors to start that way.

In sum, I think you have plenty of experience with nipype that allows you to advance quickly even implementing things the hard way. The way you were driving this PR is preferable (IMHO).