Proper treatment of "hybrid" anatomical anchors

[dickscheid]

siibra knows some data features which have "hybrid" anatomical anchors including both a location and region specification. An example are the layer wise cell densities from BigBrain optical patches, which are assigned to a cytoarchitectonic brain area but also come with a coordinates.

siibra.features.get will then return these also for other than the specified regions, since the matching implementation will use the probabilistic maps and return the feature for any brain region which has a nonzero value at the coordinate.

This is not wrong, but confusing and probably not intended. The question is whether we should prioritize a region spec over the coordinate location if the query is itself a region. more precisely, if the region spec of the feature matches a different region from the parcellation, this could actively prevent that the location is tested.

I put this here for further discussion.