BUG: Use the appropriate index to extract b-values

[jhlegarreta]

• BUG: Use the appropriate index to extract b-values
• STYLE: Add gradient-related properties to DWI data class
• STYLE: Make bvecs/bvals serialized shapes match comments/expectations
• STYLE: Simplify getting the root name for DWI serialization
• DOC: Mark the insert_b0 parameter as optional
• ENH: Add decimal places for bval/bvec serialization

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 69.15%. Comparing base (23a1997) to head (138573f).
Report is 27 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #72      +/-   ##
==========================================
+ Coverage   68.75%   69.15%   +0.40%     
==========================================
  Files          20       20              
  Lines         957      963       +6     
  Branches      121      119       -2     
==========================================
+ Hits          658      666       +8     
  Misses        254      254              
+ Partials       45       43       -2     

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

[jhlegarreta]

A round-trip test is probably missing to prevent issues like the one in 63bdf03. Do not have the bandwidth to do that now.

[jhlegarreta]

Re: #72 (comment) There is a round-trip test in

nifreeze/test/test_data_dmri.py

Line 123 in 7322e93

assert np.allclose(dwi_h5.gradients, dwi_from_nifti1.gradients)

But there is definitely something not working or missing there that is not being captured, as otherwise the test should have failed due to the bug that this PR addresses. I cannot debug this due to issue #73.

[jhlegarreta]

Re #72 (comment): issue #79. Since fixing the issue seems to require a bit of work, including maybe some discussion or clarification as I am not convinced by the insert_b0 option or do not fully understand the rationale of other aspects of the current logic, I would leave fixing #79 for a separate PR.