[discussion] Adjust schema for transformMatrix.json

[JThoennissen]

To discuss:

1. Replace "incomingVolume" with "source space"
2. Replace "referenceVolume" with "target space"
3. Space specs should be dictionaries, e.g. for BigBrain:

   "target_space": {
   "@id": "minds/core/referencespace/v1.0.0/a1655b99-82f1-420f-a3c2-fe80fd4c8588"
   }

4. Specify the matrix, so it works in mm spaces, or introduce the unit separately for each space, because it is still unclear wether the input or output coordinates are expected in nm, or both. So we could do sth. like

   target_space": {
   "@id": "minds/core/referencespace/v1.0.0/a1655b99-82f1-420f-a3c2-fe80fd4c8588“,
   "unit": "nm"
   }

   And then just use "transform“: {…}

Also relates to #153

[xgui3783]

from my perspective the following will need to be done:

• [ ] [optional] update/add new entry to the json schema described at https://github.com/FZJ-INM1-BDA/voluba/blob/fa8e58d/.openminds/voluba.v1.linearTransform.json and increment the file version
• [ ] allow voluba to export in the said format
• [ ] allow voluba to import in the said format, but allow backwards compatibility

[JThoennissen]

Comments by @pacher:

• depends on the use cases, namely what software is expected to work with transformMatrix.json

1. voluba / siibra:

• leave it in neuroglancer format (nm)
• conversion to other units only for storage, will increase complexity and will be a source of bugs besides extra work

2. other software:

• see if we can provide "export" or "convert" functionality to convert to a different format expected by the tool
• like the image service can export nifti with updated affine given the matrix in neuroglancer form
• it is not only the units

Suggestion:

• if we have a proper reason and valid use case: change it
• if we want to be more consistent: adapt some kind of standard (even if it is nifti afiine matrix)
• only changing units will just confuse everybody