Marking a BOLD/ASL series as slice-timing corrected

[effigies]

This is related to #836, but addresses the more specific issue of how to determine whether a BOLD series (or other 4D series) has been slice-timing corrected.

I have discussed two options in the last couple days:

Metadata key	Data type	Description
SliceTimingCorrected	bool	Slice-timing correction has been applied to the file. Any temporal shift induced MUST be accounted for in StartTime.
SliceTiming	array of numbers	[Current description +] If slice-timing correction has been applied to the file, then all values MUST be the same. The acquisition time of the volume will be interpreted to be StartTime + RepetitionTime * vol_idx + SliceTiming[0].

Or similar. The former would clearly go under imaging derivatives. The latter I would propose be directly added to the SliceTiming definition, although it would only be invoked in derivative datasets.

I'm a bit wary of this latter, as this would complicate the specification of sparse acquisition sequences (with gaps between successive volumes). Currently these can use SliceTiming to determine the acquisition time (TA). But if that's the way people want to go, we can work out the consequences.

[Remi-Gau]

I have a preference for the former: it is more transparent for users without having to go and dig into the specs.

Also if there one thing that we do NOT need more of in the specs is another way to express slice timing. 😉

Once upon the time I added a table to explain the "mutually exclusiveness" of the different slice timing options, and even now with the table I have to go through the logic of it when I read it.

So unless we want to make @tsalo 's favorite example of the headache of handling conditional logic with the BIDS schema even more complicated, I would say SliceTimingCorrected sounds great.