Reconsider output hierarchy order

[Michael Waskom]

Related to #110:

If the idea is to define a single functional template space across experiments, it might make more sense for the parameterization to be:

analysis_dir
  |
  <subj>
    |
    <template>
    |
    <exp_a>
      |
      timeseries
      |
      <model a>

I suppose this would entail that the project-level template_exp preprocessing would need to be executed first ... but I suppose that's always the case.

A more radical idea is to have a pre-preproc template module that just operates on the spin-echo images. Not sure how to do this in a backwards compatible way for datasets without spin echo images.

Worth thinking about ...

[Michael Waskom]

Merging into #110