Disable SCORE/SCRUB for GE/short runs and get GE workflow working

[tsalo]

Closes None.

Changes proposed in this pull request

• Add BIDS-examples datasets as test data.
• Disable SCORE/SCRUB automatically for GE/short ASL data.
• Patch in rmsd as a NaN for GE data.
• Extract the M0 TR in GeReferenceFile.
• Simplify the code in GeReferenceFile, although I would still like to separate M0 extraction from the reference file generation.
• Start simplifying the SCORE/SCRUB code. It still needs a lot of work to be readable.
• Use RepetitionTime for summary interfaces when available, but resort to RepetitionTimePreparation when it's not. This probably won't work with multi-PLD data, but that isn't supported at the moment anyway.
• I still think slice-timing correction is wonky. The data are actually acquired in a very short perior within the TR, so it's not spaced evenly across the TR like BOLD scans. Additionally, we have control-label pairs where interpolating across volumes doesn't make sense. Even with deltam volumes, the TR is essentially 2X the original, and the slices come from very different times, right?