feat(dom): add safe classifiers and missing canvas bindings [Codex]

[jderochervlk]

React and DOM APIs often expose broad element types, so consumers and examples otherwise repeat raw instanceof checks, Obj.magic casts, and local property bindings.

This adds guarded, polymorphic classify helpers to Element, Document, HTMLElement, HTMLInputElement, HTMLImageElement, and HTMLCanvasElement. Each helper returns an optional concrete interface value, keeps the checked cast inside the binding, and returns None when the corresponding constructor is unavailable.

The test-backed DOM examples now use the public classifiers directly. The canvas example also uses CanvasRenderingContext2D bindings for fillStyle, font, and textBaseline instead of defining those bindings locally. The generated API comments and DOM guidance explain when narrowing is needed, option handling, unavailable constructors, and the current-realm limitation.

Runtime coverage exercises successful classifications, inherited interface matches, mismatches, and missing global constructors. No dependencies are added.

[Freddy03h]

I think this is essential, even if it means having some runtime code instead of keeping everything zero-cost.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-18T14:01:02.026206Z	670bba7	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[Freddy03h]

Actually, we should update the DOM examples and also look at other examples that implement bindings or conversion helpers, and add those to the library as well.

[jderochervlk]

Followed up in 670bba7: the input and canvas examples now use public classifiers, HTMLCanvasElement and HTMLImageElement have guarded classifiers, and the canvas property helpers live in CanvasRenderingContext2D. I also audited the other examples; the remaining local externals and Obj.magic values are test fixtures rather than reusable binding gaps. Verified with npm test, npm run format:check, all 27 feature builds, and the full docs build.