Unit overhaul 3/6: Store Q as a scipp Variable in 1/angstrom

[henrikjacobsenfys]

Change _validate_and_convert_Q to return a scipp Variable (dimension 'Q', unit 1/angstrom) instead of a bare numpy array, and update the model hierarchy accordingly: ModelBase, InstrumentModel and DiffusionModelBase now expose Q as a Variable, compare Q with sc.allclose, and iterate via range(len(Q)); DeltaLorentz and the translational-diffusion models take .values where they need raw floats for naming and math.

This carries the canonical Q unit through the public .Q property, so scipp Q values in any convertible unit (e.g. 1/nm) are accepted and normalized. Tests updated to read Q via .values.

All unit tests pass.

---

Stack — splitting the monolithic "Unit system overhaul" (#217) into a reviewable series. Merge bottom-up:

1. Add unit-conversion utilities and FitTarget → develop (Unit overhaul 1/6: Add unit-conversion utilities and FitTarget #221)
2. Centralize Q-index validation via verify_Q_index → split/01-unit-utils (Unit overhaul 2/6: Centralize Q-index validation via verify_Q_index #222)
3. Store Q as a scipp Variable in 1/angstrom → split/02a-verify-q-index (Unit overhaul 3/6: Store Q as a scipp Variable in 1/angstrom #223) (this PR)
4. Separate model unit into x_unit/y_unit + update tutorials → split/02b-q-variable (Unit overhaul 4/6: Separate unit into x_unit/y_unit and update tutorials #224)

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.02%. Comparing base (9abb18f) to head (c467b31).
⚠️ Report is 2 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop     #223   +/-   ##
========================================
  Coverage    98.01%   98.02%           
========================================
  Files           53       53           
  Lines         3685     3687    +2     
  Branches       640      640           
========================================
+ Hits          3612     3614    +2     
  Misses          43       43           
  Partials        30       30           

Flag	Coverage Δ
unittests	98.02% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
...iffusion_model/brownian_translational_diffusion.py	100.00% <100.00%> (ø)
...mics/sample_model/diffusion_model/delta_lorentz.py	97.58% <100.00%> (+<0.01%)	⬆️
...mple_model/diffusion_model/diffusion_model_base.py	96.87% <100.00%> (ø)
...el/diffusion_model/jump_translational_diffusion.py	100.00% <100.00%> (ø)
src/easydynamics/sample_model/instrument_model.py	100.00% <100.00%> (ø)
src/easydynamics/sample_model/model_base.py	97.56% <100.00%> (ø)
src/easydynamics/utils/utils.py	98.86% <100.00%> (ø)

🚀 New features to boost your workflow:

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

[AndrewSazonov]

LGTM

Non-blocking: it looks like Q now has two different meanings in different parts of the code: sometimes it is a unit-aware scipp.Variable, and sometimes it is a raw NumPy array of values. Assignments like Q = Q.values make that switch less obvious. Consider using a separate name, for example q_values = Q.values, in places where the code intentionally drops the unit and works with raw values.