Handle characteristic-divisible fractional elliptic-curve morphisms

[harrshita123]

Fixes #42576.

Problem

EllipticCurveHom_fractional.scaling_factor() divided the numerator's scaling factor by the denominator in the base field. When the characteristic divided the denominator, both values could become zero, producing an undefined 0/0.

The same case exposed two related gaps in to_isogeny_chain(): the quotient's inseparable degree was computed but not included in the resulting chain, and the ordinary characteristic-(p) kernel calculation retained too much (p)-torsion.

Changes

• Build the quotient's inseparable factor explicitly with the appropriate Frobenius isogeny.
• Transport geometric kernel generators to the Frobenius twist before constructing the remaining separable chain.
• Correct the ordinary characteristic-(p) kernel generator after division by a (p)-power.
• Determine the final post-isomorphism by evaluation when differential scaling cannot distinguish it.
• Obtain the scaling factor from the explicit chain when the denominator is zero in the base field.
• Add regression tests for non-scalar, supersingular, and still-inseparable quotients.

Validation

• ruff check --config .github/workflows/ruff.toml src/sage/schemes/elliptic_curves/hom_fractional.py
• python3 -m py_compile src/sage/schemes/elliptic_curves/hom_fractional.py
• git diff --check

The checkout used for development does not contain a built Sage runtime, so the full Sage doctest command could not run locally. The added doctests are included for CI validation.

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation and checked the documentation preview.

⌛ Dependencies

None.

[github-actions]

Documentation preview for this PR (built with commit f36f2d9; changes) is ready! 🎉
This preview will update shortly after each push to this PR.