Zeiss CZI: fix assembly of overlapping tiles

[melissalinkert]

Fixes https://trac.openmicroscopy.org/ome/ticket/12843.

To test, import the file from zeiss-czi/sebastian/12843 into OMERO; wait for the pyramid to be generated, then compare the resulting image with what is shown in Zeiss' Zen software.

[ghost]

Testing on trout; still generating pyramids after 6 hours. This looks like it might be a more general OMERO regression.

[melissalinkert]

Pushed one more commit to fix test failures noted in https://ci.openmicroscopy.org/view/Bio-Formats/job/BIOFORMATS-5.1-merge-full-repository/485/

[mtbc]

We have a Travis failure.

[melissalinkert]

Restarted.

[ghost]

Pyramids are now being generated and they do match what I see in Zen Blue. Incidentally, do we actually need to generate pyramids or does the CZI already contain them? Over a network drive, Zen loaded the zoomed-out view and smoothly zoomed in and out across the whole image, making it look like it already had the data to hand.

Looks good to merge if the jobs are all green now.

[melissalinkert]

Using the pyramids may be possible in the future, but is outside the scope of this PR.