Limit on timeseries length

[edeguzman]

At the moment the code is limited in the length of timeseries that can be preprocessed. Here are two places I have noticed the problems crop up:

1. Change TR: The timeseries can not have more than 9999 points, or the step that puts the timeseries back together will do it in the wrong order. rsfmri_preprocess_02_change_tr.sh
2. Regress nuisance: Code crashes if the number of voxels is less than the number of timepoints. This is a problem for 2D data. The kluge workaround at the moment is to pad the data in space (along z) before running this step rsfmri_preprocess_08_regress_nuisance.sh