ENH: Add a shell data property to DWI data class

[jhlegarreta]

Add a shell data property to DWI data class that returns a list of pairs consisting of the estimated b-value and the associated DWI data.

Closes #66.

[jhlegarreta]

Opened as draft as I cannot debug this locally due to issue #73.

Edit: There seems to be some data formatting to be done to compare the DWI data obtained to the expected data, but the approach seems to be working otherwise:
https://github.com/nipreps/nifreeze/actions/runs/12969669089/job/36173994765?pr=74#step:12:1064
https://github.com/nipreps/nifreeze/pull/74/files#diff-f311d147ca2f3984f30068c37db4ca0a780e387581219cd6fa50a33e917d12cbR204

Edit: managed to debug locally undoing the type hinting related stuff. Ready to be reviewed.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 69.28%. Comparing base (628aa8f) to head (a3a511f).
Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #74      +/-   ##
==========================================
+ Coverage   69.15%   69.28%   +0.12%     
==========================================
  Files          20       20              
  Lines         963      967       +4     
  Branches      119      119              
==========================================
+ Hits          666      670       +4     
  Misses        254      254              
  Partials       43       43              

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jhlegarreta]

Maybe it is worthwhile thinking about the interplay between this feature and the b-value iterators, and how the whole thing fits into the estimators' pipelining philosophy.

[oesteban]

Maybe it is worthwhile thinking about the interplay between this feature and the b-value iterators, and how the whole thing fits into the estimators' pipelining philosophy.

I agree. In our last iteration we made the iterators independent of the object, so now you could go shell by shell and run any available iterator (those without concept of b-value, evidently).