Retain Python 3.14 scalar and Bingham ownership fixes

[FlorianPfaff]

Summary

This branch was manually rebased onto the now-green main after its original 20-commit CI-repair stack conflicted with #4293.

Already-landed or obsolete changes were dropped, including the diagnostic workflow, circular integration repair, PyTorch axis compatibility changes, wrapped-normal order handling, and the coarse-grid assertion adjustment.

The two unique fixes were retained:

• accept zero-dimensional and length-one backend arrays as scalar von Mises parameters, including the SciPy sampling path on Python 3.14
• use the active backend's owning copy in BinghamFilter._conjugate() so PyTorch inputs are not mutated

Regression coverage

• length-one von Mises parameters remain accepted by set_mean() and sample()
• Bingham quaternion conjugation returns the expected value without modifying its input

Rebase result

• based directly on current main
• 4 commits ahead, 0 behind
• 4 focused files changed

The full backend matrix is running on the rebased branch.

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no	no	27.61s
✅ JSON	prettier	7	0	0	0	1.78s
✅ JSON	v8r	7		0	0	4.93s
✅ MARKDOWN	markdownlint	68	0	0	0	1.94s
✅ MARKDOWN	markdown-table-formatter	68	0	0	0	0.8s
✅ PYTHON	black	1491	182	0	0	94.96s
✅ PYTHON	isort	1491	329	0	0	2.66s
✅ REPOSITORY	betterleaks	yes		no	no	2.36s
✅ REPOSITORY	checkov	yes		no	no	55.4s
✅ REPOSITORY	gitleaks	yes		no	no	16.0s
✅ REPOSITORY	git_diff	yes		no	no	0.32s
✅ REPOSITORY	secretlint	yes		no	no	58.77s
✅ REPOSITORY	syft	yes		no	no	6.22s
✅ REPOSITORY	trivy-sbom	yes		no	no	6.92s
✅ REPOSITORY	trufflehog	yes		no	no	33.1s
✅ YAML	prettier	11	0	0	0	0.72s
✅ YAML	v8r	11		0	0	12.55s
✅ YAML	yamllint	11		0	0	0.54s

Notices

📣 MegaLinter 9.5.0 is out! Discover the new features and security recommendations in the release announcement. (Skip this info by defining SECURITY_SUGGESTIONS: false)

See detailed reports in MegaLinter artifacts

Your project could benefit from a custom flavor, which would allow you to run only the linters you need, and thus improve runtime performances. (Skip this info by defining FLAVOR_SUGGESTIONS: false)

• Documentation: Custom Flavors
• Command: npx mega-linter-runner@9.6.0 --custom-flavor-setup --custom-flavor-linters PYTHON_BLACK,PYTHON_ISORT,COPYPASTE_JSCPD,JSON_V8R,JSON_PRETTIER,MARKDOWN_MARKDOWNLINT,MARKDOWN_MARKDOWN_TABLE_FORMATTER,REPOSITORY_CHECKOV,REPOSITORY_GIT_DIFF,REPOSITORY_GITLEAKS,REPOSITORY_BETTERLEAKS,REPOSITORY_SECRETLINT,REPOSITORY_SYFT,REPOSITORY_TRIVY_SBOM,REPOSITORY_TRUFFLEHOG,YAML_PRETTIER,YAML_YAMLLINT,YAML_V8R

Show us your support by starring ⭐ the repository

[FlorianPfaff]

Automatic rebase onto current main was attempted after #4293 landed, but GitHub reported merge conflicts. This branch is an older 20-commit CI-repair series that overlaps the newly merged CI fixes in backend signal handling, PyTorch compatibility hooks, circular distributions, and related tests. I left the original branch unchanged. It is currently 20 commits ahead and 15 commits behind main; it needs manual conflict resolution or can be superseded by #4293.

[FlorianPfaff]

Resolved manually after the automatic rebase conflict. The branch is now based directly on current main (4 commits ahead, 0 behind). Obsolete diagnostic-workflow changes and CI fixes already landed via #4293 were dropped; the two unique fixes were retained: Python 3.14-safe length-one von Mises scalar handling and backend-safe Bingham conjugation, with their focused regressions.