Fix: scale the operator when dividing it by a scalar

[raashish1601]

Op / y solves y = Op x, but LinearOperator.__truediv__ didn't check whether y is a scalar. So Op / np.sqrt(2) ran a solve against a constant data vector and returned an array, while Op / 2.0 failed deep inside cgls or lstsq (#803).

This takes option 1 from the issue: a scalar divisor (anything with np.ndim(y) == 0, so Python, NumPy and 0-d array scalars) now returns self * (1 / y), the same scaled operator Op * a gives. Non-scalar y still goes through the solve as before.

The FFT, FFT2D and FFTND operators override __truediv__ with their exact inverse, so they had the same problem (FFT(dims=8) / 2.0 raised AttributeError). Each of those overrides now hands a scalar to the base class first. I also added a sentence to the div docstring.

Tests: test_scalar_division in pytests/test_linearoperator.py (an explicit MatrixMult and a non-explicit VStack, real and complex, square and overdetermined) and in pytests/test_ffts.py (FFT, FFT2D and FFTND for each engine, skipping fftw and mkl_fft when they aren't installed). Both fail on dev and pass here. pytests/test_ffts.py, test_linearoperator.py and test_solver.py pass locally (fftw/mkl_fft not installed here), and ruff check / ruff format are clean on the changed files.

AI assistance: this PR was prepared with an AI coding agent (Claude Code, Anthropic). It reproduced the issue, wrote the fix and tests, and ran them locally as described above.

Closes #803

[codacy-production]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

View in Codacy

🟢 Metrics 12 complexity · 19 duplication

Metric	Results
Complexity	12
Duplication	19

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}

[mrava87]

@raashish1601 thanks for tackling this issue.

Overall the fix is good and clean... I just left a few comments, mostly things that did not match with other common patterns in the codebase, please fix them before I can approve and merge 😄

[mrava87]

@raashish1601 thanks for the changes!

I just left two minor comments, after they are addressed this is ready to be merged 😄