feat(cctv): seven camera packs, semantic labels, terrain-clear monitor planes (#433)

Seven contributed CCTV camera packs (Ontario 511, Fintraffic Finland, DriveBC, TxDOT Texas, Estonia Tallinn + Tarktee, Warendorf, Live Traffic NSW) on a fair per-pack catalog cap; semantic camera labels; monitor planes lifted clear of the terrain from a precomputed, bundled ground-height dataset. Default catalog 800 → 3,446 cameras. Full context in PR #433.

Co-authored-by: sendmebits <[email protected]>
Co-authored-by: avocadoAgent <[email protected]>
Co-authored-by: Alex Harvey <[email protected]>
Co-authored-by: Jason Meredith <[email protected]>
Co-authored-by: Daniel Smith <[email protected]>
Co-authored-by: Vladimir Sobolev <[email protected]>
Co-authored-by: Matthias Schneider <[email protected]>
Co-authored-by: Cyber Knight <[email protected]>
Co-authored-by: NovaSoft Australia <[email protected]>

src/layers/cctv/selection.js

@@ -135,6 +135,10 @@ export function createSelection({
     // until update()'s one-shot tiles-ready completion pass re-grounds it.
     parts.ground.rearmGroundResolution(record);
     parts.geometry.updateRecordGeometry(record);
+    // Ground under the plane's footprint (DEM, cached proxy) for cameras
+    // without a shipped precompute for this pose — only the camera the user
+    // is looking at ever asks.
+    void parts.ground.resolveFootprintGround(record);
     record.activationDone = true;
     parts.rendering.refreshCoverageStyles();
     // The newly active camera leaves the ambient ring (its monitor plane takes