Add modulo support for LI-FLIM data

[melissalinkert]

Fixes http://trac.openmicroscopy.org.uk/ome/ticket/12290. Phase and frequency dimensions in LI-FLIM files are now stored in the appropriate Modulo annotation, rather than being stored as separate series/Images.

To test, verify that QA 9236 has two Images when imported - the first should contain 12 planes (for 12 phases), and the second should contain a single plane (the background image).

I would suggest also checking other datasets from the data_repo/from_skyking/li-flim directory, as many of those files are also multi-phase/multi-frequency datasets which have now had their import behavior changed. In all cases, I would expect no more than 2 Images to appear after import.

[joshmoore]

Relaunched travis (connection reset to maven central)

[mtbc]

One query: for from_skyking/li-flim/samples/sample-extended.fli the primary image looks in Insight to have 12 × Z, 5 × T, 12 × t where, for constant T and t, the different Z all look the same. (The background image's 12 × Z look different.) Is that okay?

The 12 planes of QA 9236's primary image come through as 12 × t, so apart from the above query this looks good to merge.

[melissalinkert]

If I open sample-extended.fli in ImageJ, the Z sections are definitely distinct. You might need to zoom in and fiddle with the rendering settings though - there are no identifiable objects in these images and the min/max range is large enough that it's easy not to spot the difference between planes.

[mtbc]

@pwalczysko, could you see if you can see a difference on Z for constant T, t? The image is on trout under read-only-1 user-2 in dataset mtbc-pr-1164 as image ID 33551.

(@melissalinkert: I don't know if the CI jobs already reassure that there has been no regression here.)

[pwalczysko]

Looks like an Insight bug. @mtbc is filing a ticket, thanks.

[mtbc]

So, this PR is good to merge.

[melissalinkert]

--rebased-to #1197