Create synthetic exemplar datasets

[Lestropie]

See for example proposed changeset in bids-standard/bids-specification#1003:
https://github.com/bids-standard/bids-specification/pull/1003/files#diff-926444c20e69f74c364311f3c798e0b7d905f30d4bd93ca7b087372ad2fcf965R91-R222.

This project would require creation of a synthetic dataset that is tailored for both demonstration of the Inheritance Principle (additionally under different existing or proposed rule sets) and proof of correct operation of the software.

Ideally this dataset would include multiple subjects, sessions, modalities, runs, etc. etc.. For each (zero-sized) data file, add metadata fields into a corresponding sidecar, which contain information that would be expected to be consistent across some experimental dimension (eg. multiple runs of the same modality), but differ across others (eg. acquisition time of different sequences), and be present in some but absent in others (eg. magnetic field strength being applicable to MRI data only).

As software implementations for exploitation of the Inheritance Principle are realised for different rule sets (eg. #2, #3, #4), the outcomes of those operations would be added to the exemplar dataset. These could then be used for:

• The communication of what the Inheritance Principle might look like for different proposals for either BIDS 1.x or BIDS 2.0.
• Validation of other BIDS ecosystem software packages, ensuring that both discovering all applicable metadata files for a given data file, and discovering all data files to which a given metadata file is applicable, operate as expected under a given Inheritance Principle rule set.

[Lestropie]

Whether within a singular dataset, or potentially as a separate dataset, there should exit situations where a subset of metadata are overridden, as is permitted in BIDS 1.x. This can be used to:

• Verify that the internal handling of the Inheritance Principle appropriately overwrites the metadata contents only for those data files where such overloading should occur.
• Verify that converting this dataset into one where the Inheritance Principle is used, but overriding is not permitted, yields a dataset where the interpretation of metadata remains consistent with the original dataset.
  (In this way, the software tool could be used to remove the use of metadata overloading where it has previously been used, but without removing use of the Inheritance Principle altogether)