Cache load3DModel results

[Jam-Ngai]

Summary

• Add a shared cache for load3DModel so duplicate model URLs share the same in-flight or completed load.
• Normalize cachebust_origin out of cache keys while preserving the original URL passed to Three loaders.
• Detect model file extensions before query strings or hashes, so cache-busted .obj, .stl, .wrl, .gltf, and .glb URLs still select the correct loader.
• Return cloned model instances with cloned materials so per-component transforms, hover state, and translucency changes stay isolated.
• Add focused regression coverage for cache-key normalization, query/hash extension detection, material cloning, and concurrent deduplication.

Verification

• git diff --check

I could not run the Bun/TypeScript test suite locally because this environment does not have bun, npm, tsc, or biome installed, and the Homebrew GitHub download path was timing out earlier.

AI-assisted with Codex and manually reviewed.

/claim #93

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
3d-viewer	Ready	Preview, Comment	May 17, 2026 6:54am