Fix export of multiple files

[melissalinkert]

See https://trac.openmicroscopy.org.uk/ome/ticket/10635

This needs manual testing, ideally both in ImageJ and using bfconvert. All stack output formats (OME-TIFF, TIFF, QuickTime, AVI, PNG, ICS) are affected; readers are not affected at all.

In ImageJ, best would be to test opening a single dataset with SizeZ, SizeC, and SizeT all greater than 1, then exporting to each of the above formats with one or more of the "Write each ..." boxes checked. No error message should be shown, and the converted files should all open correctly.

With bfconvert, just testing that bfconvert input-file output-file works for each of the above output formats should be enough.

[ghost]

Export of avi split by c+t ran out of file handles Over 10000 open after doing a few exports of different formats (1400 planes/image). Possible leak?

See https://gist.github.com/rleigh-dundee/5445697

(Ran out of filehandles for copying it!)

[ghost]

Export of multiple images works correctly for both bfconvert and imagej. OME-TIFF tested for all possible combinations of splitting, and all reimport correctly. However, all writers appear to leave all filehandles open.

[melissalinkert]

Open file handle issues should be fixed now.

Only other issue I know of in connection with this PR was with /ome/team/rleigh/ics-test.zip. In checking, the reported behavior (picking any file with "Group files with similar names" checked only opens one timepoint) is correct, or at least expected. Each timepoint consists of multiple files, presumably because the .ids extension was used when converting. File grouping is disabled when the dataset already has multiple files; exporting with the .ics extension instead would solve the problem.

[ghost]

No file handle leaking now observed (testing with FiJi). Please merge.