DICOM: try to handle Siemens mosaic even if CSA fields missing

[jdtournier]

see related discussion in #1012.

Is this worth considering including? Slight problem in that it requires a bit of work to finalise in order to throw out redundant slices, but should still be do-able relatively quickly...

[thijsdhollander]

Since not finalised yet, I propose to save it for a future release, or a hotfix.

[Lestropie]

@jdtournier: Did my best to resolve this with dev, but it will need your look-over / a run with your DICOM test suite / a test with scrubbed CSAs.

Old review from RC2 preparation days no longer relevant.

[thijsdhollander]

Is this still a goal for RC3?

[jdtournier]

Is this still a goal for RC3?

To be honest, I don't think it should be. It certainly shouldn't delay the release, and even if I do get around to implementing the remaining changes, it'll require testing to make sure nothing breaks. I'll give it a crack if I have a minute, but personally I think we should release ASAP regardless.

[Lestropie]

If it requires any more than just running your DICOM import test script, then let's leave it out.

[jdtournier]

OK, let's leave it out.

For completeness, the testing I'd do is just to make sure it doesn't break existing well-formed data. I'd need additional testing to check that it can read malformed data correctly, but there's no reference data for that to test against. So yes, it's a bit of work...

[Lestropie]

@jdtournier Will leave in 3.1.0 milestone pending a re-run through the DICOM test suite.