ENH: Refactor data representation structures for multimodality

[oesteban]

Proposes a new nifreeze.data.base.BaseDataset class with minimal metadata shared across modalities. Modality-wise data representation classes should derive from this one and extend the necessary behaviors (e.g., indexed access).

I believe this approach would be cleaner to have a common interface. Models can be more specific in the required data structures they handle.

By adding indexing, the dataset can be split with fancy indexing, but returning a tuple containing the split across data AND metadata. This will allow transparent splits within models.

Resolves: #19.

[codecov]

Codecov Report

Attention: Patch coverage is 71.68142% with 64 lines in your changes missing coverage. Please review.

Project coverage is 67.60%. Comparing base (fdab87b) to head (68ccc85).
Report is 14 commits behind head on main.

Files with missing lines	Patch %	Lines
src/nifreeze/data/pet.py	24.00%	38 Missing ⚠️
src/nifreeze/data/base.py	86.17%	8 Missing and 5 partials ⚠️
src/nifreeze/data/dmri.py	87.17%	4 Missing and 6 partials ⚠️
src/nifreeze/registration/ants.py	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #52      +/-   ##
==========================================
+ Coverage   65.85%   67.60%   +1.75%     
==========================================
  Files          19       20       +1     
  Lines         943      991      +48     
  Branches      121      131      +10     
==========================================
+ Hits          621      670      +49     
+ Misses        277      267      -10     
- Partials       45       54       +9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jhlegarreta]

Approving for the sake of moving forward: comments can be addressed in a separate PR.

So this means I will need to fix all conflicts in PR #28 😭.

Thanks for this, thanks for the notebook.

[arokem]

Just a couple of thoughts about whether we want something more generic for the stored affines. But I have not thought this through very deeply.

[oesteban]

So this means I will need to fix all conflicts in PR #28 😭.

See 68ccc85 to minimize those conflicts.

[jhlegarreta]

See 68ccc85 to minimize those conflicts.

Thanks.