configuration for automated "publishing" of collected datasets

[yarikoptic]

we might want to publish collected datasets to another host for a number of reasons, e.g.

1. this box is not accessible by the group and only serves for conversions, so we might want to publish somewhere else with all the data
2. we might want to publish datasets to some publicly available web site without actual data (as we do now already but manually ;) )
3. in some kind of ideal universe, we publish all sanitized (defaced etc) data right away for public consumption. Actually could be setup for QA sub-dataset(s) right from the beginning
4. where desired, we could even setup per PI dataset remotes, to push where allowed directly to their lab-servers

To accomplish this we could rely on inheriting configuration ('datalad publish --inherit', which will be there in datalad 0.5) but then we should just have an option to heudiconv (e.g. --datalad-publish) with a list of remotes to publish to (or may be to all?) while relying on manually establishing remotes first within the base/target dataset. It would require additional pass down to the dataset to have 4. effect (since that remote would not be present in the topmost super dataset)