Sort data repository into different categories

[curtis]

Our repository of test files contains a large quantity of data: over 140 GB. We would like Hudson to run the automated Bio-Formats TestNG suite every time we do a commit, but 140 GB is far too much data. We need to split the repository into the following subsets:

1. Publically available sample data, posted on the web site; ideally, at least one dataset in each supported format, assuming we can obtain permission
2. Data we auto-test on each commit; at least 1-2 datasets in each format, including all publicly available datasets
3. Data we auto-test before each release; a larger number of datasets in each format
4. Other/auxiliary data, not part of the automated test suite

[jmoore]

imported from ​bio-formats:#308

[mlinkert]

(1) has already been done, (3) is covered by the "Bio-Formats full repository test" on LOCI Hudson (i.e. the full repository is split into seven pieces, so that everything is tested in the space of a week). This renders (4) irrelevant, leaving just (2) to be done.

[mlinkert]

There is now a 'test_per_commit' directory in the repository that covers (2).