Bug: ZVI ROI parsing issue

[rleigh]

zeiss-zvi/pki/E-07.zvi:

Found 0 ROIs, but 121 ROIs expected

Looks like we're not processing the ROI layer block properly for this file.

[mlinkert]

See also:

​https://list.nih.gov/cgi-bin/wa.exe?A2=ind1304&L=IMAGEJ&F=&S=&P=99821

[rleigh]

For E-07.zvi:

00000000  03 00 04 10 00 10 08 00  08 00 00 00 6b 00 65 00  |............k.e.|
00000010  79 00 00 00 03 00 01 00  02 00 03 00 00 00 00 00  |y...............|
00000020

The ROI block is non-standard, so we erroneously think that there are 121 (0x79) ROIs when there are none. This is because there's a standard layer version "04 10 00 10", but we normally expect this to be followed immediately by a shape. Here the shape count after the "03 00 01 00 02 00 03 00" is actually 0. The additional bit is a string "key" ("08 00" with zero length, but null terminated). This can probably be special cased, but is otherwise harmless. Not clear what this is for, maybe it's the layer name. Why is the string length missing?

​https://github.com/openmicroscopy/bioformats/pull/471 fixes the real parsing bugs found in the reader. The cosmetic issue above will be done next.

[rleigh]

Fixed and merged.