ENH: add capability to concatenate multiple runs per subject

[mwvoss]

Summary

For some participants we may have multiple resting state scans and ideally our estimated fc values are across concatenated scans rather than the average of separate scans.

Actual behavior

Scripts assume one scan per subject

Expected behavior

It would be nice to add a concatenation option.

Script/Workflow details

The main issue is that we're seeding in native space so we need to be careful with the 4D residual volume to ensure we're aligning them properly before extracting a seed timeseries from the concatenated volume. @brussj had adapted some code for this that we could incorporate:

1. post-bandpass motion scrub res4d file
2. push all EPI masks to MNI, combine into one and look for maximal overlap. This mask will then be applied to all EPI runs
3. motion scrubbed res4d files are pushed to MNI space, merged
4. Merged file is norm and scaled, then masked by maximal binary mask intersection from step partial restructuring of DataPrep #2

All the transforms should exist, it's just add the option downstream from each nuisance regression. We won't be using this option in the near future so it's not high priority, but it may come up.