Wrong formatting in BOLD json files

The information in the BOLD json files does not follow BIDS. Instead of using keys specified in BIDS it looks like the DICOM header names were used directly. This may create problems further processing this data, especially since it concerns a multi-echo dataset.

From the BIDS specification RepetitionTime and Task Name are the only required fields. For single echo data fMRIprep only seems to use the Repetition Time, Task Name, and SliceTiming information. It is also possible to run without slice timing correction, but this might not be what we want.

The SliceTiming key, necessary for slice timing correction, is not defined. The information of slice timings might still be available in the MosaicRefAcqTimes (see discussion on spm mailing list). The values here are in milliseconds but we can tranform them before adding to a new SliceTiming key. However, sometimes these times are given in reverse order. This can normally be determined in the SliceEncodingDirection field. This information seems to be based on the combined values of several DICOM header fields, and it is more difficult to determine if we can verify this from the information we have. @richlan Do you know anything about getting this information from the dicom header?

Additionally, for processing multi-echo data, we need the EchoTime. This key exist but value looks like it is in miliseconds again so we would have to transform it.

We would still have to verify if this is enough to process the data, so let's decide if it is worth it to attempt these fixes based on how much time it would take.