Try importing the 3 downtown bridges (SketchUp) into the Chattanooga diorama

[TortoiseWolfe]

Goal

Try replacing the 3 downtown Tennessee River bridge massing boxes in the Chattanooga diorama with real 3D bridge models. Source files are in the repo root: 1.skp, 1+final+3.skp, 2.skp (SketchUp). These correspond to the Walnut St (pedestrian), Market St (Chief John Ross), and Veterans (Olgiati) bridges — walnut_st_bridge is already a hero in sites/chatt.json.

Approach — reuse the existing multi-model pipeline

The general path already exists (scripts/warehouse/* + src/world/WarehouseModels.tsx + loadWarehouseModels() → twins/<slug>/models/models.json). Bridges go there, NOT the single-scan "house" path.

• Conversion (first, and the feasibility risk): SketchUp .skp is proprietary and can't be loaded headlessly by three.js. Options: (a) owner re-exports each bridge from SketchUp as GLTF/OBJ/DAE (like the Polycam DAE for the house), then convert to GLB via the existing scripts/house/convert-scan.mjs (DAE→GLB) or a GLTF passthrough; (b) find/stand up a headless .skp→glb converter (Blender + a SketchUp importer, or the SketchUp SDK). Decide before building anything downstream.
• Optimize the GLBs (scripts/warehouse/abstract-glb.mjs or scripts/house/optimize-scan.mjs).
• Add models.json entries geo-anchored by lat/lon, reusing the geocoded-placement just built for the house twin (place by true coordinates, project into the render frame, center the model on its anchor).
• Render via WarehouseModels; hide the 3 bridges' OSM massing boxes (like house.hideBuildingIds).
• Verify in /chatt?diorama (wide) that the 3 bridges sit on the river at their real spans.

Notes

Aesthetic: the diorama is a toy tilt-shift look — detailed bridge models may need simplification to fit. This supersedes the earlier "drop all SketchUp" decision, which was about the cluttered abstracted-landmark set, not curated hero bridges.