Imspector metadata fixes

[melissalinkert]

This fixes several metadata parsing bugs in ImspectorReader that affect the dimensions and ModuloAlongT. The following tickets should be fixed with these changes:

https://trac.openmicroscopy.org/ome/ticket/13002
https://trac.openmicroscopy.org/ome/ticket/13180
https://trac.openmicroscopy.org/ome/ticket/13181
https://trac.openmicroscopy.org/ome/ticket/13182

Existing Imspector .msr files should be unaffected, but new files for QA 17102 and QA 11369 need to be tested. Ticket 13180 references a file in QA 17101, but the same file is also present in QA 17102.

For each file, check showinf -nopix -omexml and verify that the step in ModuloAlongT matches the time per FLIM channel referenced in each ticket (usually 220.875 ps). If an expected value is not in the tickets, check that end in ModuloAlongT is greater than 1.

Also verify that SizeZ, SizeC, and SizeT match the expected values in the tickets. The images will be easier to check in ImageJ; in particular check that the second channel looks plausible (for files with SizeC == 2) and that the plane ordering seems sensible (i.e. the dimension order is not mixed up).

This should not affect memo files, so could be deferred until 5.6.1 if necessary. If @imunro gets a chance, a quick review from the original reporter/triager of the above tickets might be useful.

[imunro]

@melissalinkert @sbesson Thanks for looking at this. When I last spoke to the LaVision dev team they warned us that this format is subject to constant undocumented revision. We have concentrated since then on supporting their variant on ome-tiff (not without it's own problems) My apologies I should have noted this in the tickets.

[sbesson]

Discussed earlier today. Leaving this out of 5.6.0 and we can review in 5.6.1.

[imunro]

I'll try and get the FLIMfit builds running again so we can ,potentially, send a test version (inc these changes) to our LaVision users.

[sbesson]

@imunro: any success with #2913 (comment), should we try to coordinate on the review of this PR?

[imunro]

From my POV the only way I can test this PR is to build a version of FLIMfit incorporating these changes and send it to our known LaVision users to try with their data.
I'm currently working to get FLIMfit back to a status where we can do a release to have a tested/working base to start from. Sorry to be so slow but nobody is actually employed on this.
Once we're there (hopefully by End Aug) I can try this - sorry to be so slow.

[dgault]

The code changes here look good and builds and tests are green.
The existing repo files appear to be unchanged.

Testing the related QA files:

QA 11369
01 - 770nm FITC - HybridPMT.msr opens and displays correctly. The step value in ModuloAlongT is 54.98 and sizeT is 237, these values seem to be correct from what I can gather.

02 - 70 - 1 - tp1 - x16 TCSPC.msr opens and displays correctly. The step value in ModuloAlongT is 160 and sizeT is 75, these values seem to be correct.

QA 17101 and QA 17102
1 channel binning 4 time series.msr opens and displays correctly. The step value in ModuloAlongT is 110.44 and sizeT is 1298, these values seem to be correct.

1 channel binning 8 time series.msr opens and displays correctly. The step value in ModuloAlongT is 220.875 and sizeT is 649. The ticket (https://trac.openmicroscopy.org/ome/ticket/13182) mentions that there are 59 images in the series, and 11 points in the decay and 11 * 59 gives 649, so Im assuming this is the correct interpretation.

2 channels FLIM time series.msr opens and displays correctly. The step value in ModuloAlongT is 220.875 and sizeT is 295 and sizeC is 2. The step value is correct according to https://trac.openmicroscopy.org/ome/ticket/13180, Im assuming the 295 sizeT is 59 * 5 as the trac ticket mentions 5 time points.

2 channels FLIM z stack.msr opens and displays correctly. The step value in ModuloAlongT is 220.875 and sizeT is 59 and sizeC is 2 and sizeZ is 7, these values are correct according to https://trac.openmicroscopy.org/ome/ticket/13182.