Stabilize SO(3) product quaternion normalization

[FlorianPfaff]

Bug

SO3ProductDiracDistribution normalized quaternions with a direct backend Euclidean norm. For large but finite values such as [1e308, 1e308, 0, 0], the sum-of-squares intermediate overflows even though the true norm, approximately 1.4142e308, is representable.

The constructor, geodesic-distance helper, and rotation-matrix conversion could therefore reject a finite, nonzero quaternion solely because of avoidable intermediate overflow.

Fix

• reuse the existing scale-safe normalize_quaternions() helper shared by PyRecEst's other SO(3) distributions;
• retain finite/nonzero validation and canonical scalar-last quaternion handling;
• add a regression that constructs an SO(3) product Dirac distribution from extreme finite coordinates under strict overflow/invalid handling.

Validation

• reproduced the old failure locally: numpy.linalg.norm([1e308, 1e308, 0, 0]) raises FloatingPointError with overflow configured to raise;
• verified the scale-safe normalization returns [sqrt(0.5), sqrt(0.5), 0, 0] with unit norm;
• syntax-compiled the modified source and regression test;
• branch is 2 commits ahead and 0 behind main, with exactly two changed files.

The repository's pull-request workflows provide the full supported backend, Python-version, lint, and packaging validation matrix.

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no	no	16.59s
✅ JSON	prettier	7	0	0	0	0.66s
✅ JSON	v8r	7		0	0	2.79s
✅ MARKDOWN	markdownlint	68	0	0	0	0.96s
✅ MARKDOWN	markdown-table-formatter	68	0	0	0	0.37s
✅ PYTHON	black	1810	56	0	0	49.7s
✅ PYTHON	isort	1810	88	0	0	2.48s
✅ REPOSITORY	betterleaks	yes		no	no	1.11s
✅ REPOSITORY	checkov	yes		no	no	35.25s
✅ REPOSITORY	gitleaks	yes		no	no	6.48s
✅ REPOSITORY	git_diff	yes		no	no	0.1s
✅ REPOSITORY	secretlint	yes		no	no	44.65s
✅ REPOSITORY	syft	yes		no	no	3.12s
✅ REPOSITORY	trivy-sbom	yes		no	no	3.8s
✅ REPOSITORY	trufflehog	yes		no	no	21.55s
✅ YAML	prettier	11	0	0	0	0.37s
✅ YAML	v8r	11		0	0	6.56s
✅ YAML	yamllint	11		0	0	0.29s

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]

Closing because it is not merge-ready. The full test workflow fails broadly across the supported backend/Python matrix, despite the focused change looking reasonable. The normalization-helper reuse can be reconsidered as a smaller PR once it has a fully green matrix.