Create *private* RIA store

[mih]

To be put on bulk1 eventually. This will be a superset of the one from #38 and additionally contain the pristine raw data and other unreleased or unreleasable components.

Most importantly it will provide a clean separation of the "dirty" workspace on juseless, and the "managed" parts of the dataset.

Datasets to move:

• gumpdata
• 7T_ad
• pandora
• anatomy
• 3T_av_et
• 3T_visloc
• aggregate
• aligned (untracked qa/ dir, converted to proper dataset)
• annotations
• comp_stim_features
• emoreg (no result were saved, massive untracked stuff, IT WAS NO ANNEX DATASET !??!)
• fg_eyegaze (deleted, was a test for phase2)
• fg_eyegaze_raw (untracked rescue/)
• freesurfer (converted to proper dataset)
• media (untracked files)
• anondata (phase1)
• phase2
• retmap (converted to proper dataset)
• sandbox (not a dataset, connected to the timeline annotation tool)
• structural
• tnt (some untracked qa data in git-ignored dirs)
• visloc (untracked condor_logs)
• studyforrest_avmovie_shen_aggregate (deleted, contained in aggregate)
• speech-annotation

TODO:

Copy all previously existing datasets (again) with git-annex copy --all to make sure that alternative branches and previous versions remain intact!!!

Done for all datasets that are checked above, except:
datalad/datalad#5613 happening for tnt and visloc

[mih]

The store is done, but the issue linked above prevents some data from entering