TIFF: fix header writing for big-endian BigTIFF files

[melissalinkert]

Fix suggested by Matthias Baldauf:

http://lists.openmicroscopy.org.uk/pipermail/ome-users/2012-November/003463.html

[bpindelski]

1. Merged ok locally into bioformats/dev_4_4
2. Executed ant clean jars tools
3. Used tools/bfconvert -bigtiff CFPNEAT01_R3D.dv foo.tif
4. Opened foo.tif using Hex Fiend
5. First 8 bytes read: 4D 4D 00 2A 00 08 00 08

Seems to be fixed and ready to merge...

[bpindelski]

Second round of testing, after discovering that an old version of BF-tools was on the PATH. Additionally, the tested DV file was little-endian, so not fit for this test. Previous comment can be disregarded.

This time an ariol image was used and bfconvert used: ./bfconvert -bigtiff ~/Desktop/testing/import/ariol/04CBE9A7-FA1B-4556-9900-6D869E85FFC7.jpg foo.tif. The first 8 bytes of the header read: 4D 4D 00 2B 00 08 00 00, as expected.

[joshmoore]

@melissalinkert, should this have been run against any data tests via --test ?

[melissalinkert]

@joshmoore, no, --test only runs reader tests, and this is a writer-only change.

[joshmoore]

Doh. That's going to keep driving me nuts. It's a reflex.