Add the Data Library: live regional datasets, tiles and packs

DATA LAYERS keeps its curated core layers; regional and niche sources
live in the Data Library, opened from Browse data library at the foot
of the panel. Each entry is a JSON manifest in src/library/entries/
pointing at a live upstream (query API, provider-hosted file or tile
endpoint), so nothing is bundled and a self-hosted instance hosts no
data. Sources without CORS go through one allowlisted /api/library
route. Kinds: dataset (GeoJSON, CSV, JSON records, ArcGIS, partitioned
GeoParquet), tiles (XYZ, WMS/ArcGIS bbox export, terrarium relief) and
pack. Selections persist and travel in share links as lib=, separate
from the core layer tokens. Library rows get Zoom and Remove icon
chips that work while the layer is off, rendered through RailChips.
Strings are translated in every locale.

CONTRIBUTING.md

@@ -129,6 +129,7 @@ prints anything; never fix that by renaming or deleting PR A's published token.
 The highest-leverage places to jump in:
 
 - **🌆 Add a CCTV source pack.** Austin is the reference camera source. Follow the [CCTV integration workflow](docs/CCTV_INTEGRATION.md) for source review, provider registration, proxy security, tests, and attribution. The proxy only fetches server-registered URLs — never client-supplied ones; see [SECURITY.md](SECURITY.md).
+- **📚 Add a Data Library entry.** A regional or niche source that is not a core layer belongs in the Data Library: one JSON manifest pointing at a live API, provider-hosted file or tile endpoint, with its licence and attribution, and no JavaScript. See [docs/DATA-LIBRARY.md](docs/DATA-LIBRARY.md). Library entries use their own share-token ledger, not the layer tokens below.
 - **🛰️ Add or improve a data layer.** Layer factories live in `src/layers/<family>/`, with source, record, controller and renderer owners implementing the layer interface (`init/enable/disable/update/destroy/getStats`, optional `getDetectableObjects`/`getStats`). Use an existing layer as a template, and see [Adding a new layer](#adding-a-new-layer) for what the PR needs.
 - **🎙️ Extend voice control.** Voice arguments are defined in `src/voice/actionSchemas.js`, with server-side descriptions in `server/providers/openai/tools.js` and client-side execution in `src/voice/gevActions.js`. Keep the tool surface tight and the responses honest (confirm only what actually happened).
 - **🎨 Add a visual style.** Styles are GLSL post-process shaders in `src/styles/`.
@@ -228,7 +229,7 @@ ownership and adoption process.
 1. Branch off `main`.
 2. Keep `npm run build`, `npm test`, and `npm run test:track` green and avoid new console errors, plus the [feature gate](#feature-regression-gates) for the area you touched.
 3. If you change runtime behavior, update `docs/CURRENT-STATE.md` in the same PR, and add your changelog entry as `changelog.d/<pr-number>-<slug>.md` instead of editing `CHANGELOG.md` — see [changelog.d/README.md](changelog.d/README.md). Distinct filenames don't conflict; a release assembles the directory into `CHANGELOG.md`. A documentation claim that needs a regression test should assert against `docs/CURRENT-STATE.md` or `README.md`, which a PR still edits directly.
-4. If you add or change a data source, update [DATA_SOURCES.md](DATA_SOURCES.md) with its license and attribution. **Don't add data you don't have the right to redistribute** — fetch it at runtime instead.
+4. If you add or change a data source, update [DATA_SOURCES.md](DATA_SOURCES.md) with its license and attribution (for a Data Library entry, `npm run library:docs` writes its row). **Don't add data you don't have the right to redistribute** — fetch it at runtime instead.
 5. Describe what you changed and how you verified it (screenshots welcome for anything visual).
 6. If you add a new layer, fill in the new-layer section of the PR template. See [Adding a new layer](#adding-a-new-layer).