fix(cctv): harden the WSDOT catalog fetch and describe its real direction codes

The catalog fetch now refuses redirects, cancels rejected bodies and reads
through the shared capped reader before the windows-1252 decode. The feed sends
cardinal codes plus B (both), O and null, not BW or diagonals; tests, comments
and docs now say so, with unchanged id-hash fallbacks. Documents the 8-1,200
per-pack range, drops the unverified low-volume compliance claim for facts and
a link to WSDOT's terms, and stops claiming the per-pack defaults fit under
the 4,000 cap: it is filled round-robin.

server/providers/cctv/constants.js

@@ -8,8 +8,8 @@ export const DEFAULT_AUSTIN_MAX_SOURCES = 250;
  * Catalog-wide safety ceiling on served cameras. Each pack already caps
  * itself (nearest-to-anchor first); this bound only matters when the packs
  * together exceed it, and it is then filled round-robin across packs (see
- * cap.js) so no region is silently dropped. Sized above the sum of the
- * default per-pack caps so a default install never trims.
+ * cap.js) so no region is silently dropped. The default per-pack caps
+ * together can exceed it, so a default install may be thinned evenly here.
  */
 export const DEFAULT_CCTV_MAX_SOURCES = 4000;
 /** Hard upper bound for CCTV_MAX_SOURCES; also sizes the health map. */
@@ -249,9 +249,9 @@ export const VEGVESEN_IMAGE_ORIGIN =
  * per-camera manifest path is registered. */
 export const VEGVESEN_VIDEO_URL = (cameraId) =>
   `https://kamera.vegvesen.no/public/${cameraId}/manifest.m3u8`;
-/** Above the ~850 working cameras so a default install keeps them all, while
- * staying close to them: the per-pack defaults are meant to sum to no more
- * than DEFAULT_CCTV_MAX_SOURCES. */
+/** Above the ~850 working cameras so the pack keeps them all on its own. The
+ * catalog-wide DEFAULT_CCTV_MAX_SOURCES cap still applies across packs and is
+ * filled round-robin, so a full default install may thin this pack. */
 export const DEFAULT_VEGVESEN_MAX_SOURCES = 900;
 /** Hard ceiling on the catalog body. The whole country is ~800 KB. */
 export const VEGVESEN_MAX_CATALOG_BYTES = 8 * 1024 * 1024;
@@ -308,8 +308,8 @@ export const DELDOT_ANCHORS = [
 
 /** WSDOT Travel Center cameras (Washington State): keyless Esri JSON on the
  * official open-data host — ~1,700 point features in EPSG:3857 (Web Mercator),
- * ETag/Last-Modified served. WSDOT's metadata marks the feed "low volume" use,
- * which the 15-minute catalog cache respects. */
+ * ETag/Last-Modified served. WSDOT's metadata marks the feed "low volume" use;
+ * the catalog is fetched once per CCTV_SOURCE_CACHE_MS refresh. */
 export const WSDOT_CAMERAS_URL =
   'https://data.wsdot.wa.gov/travelcenter/Cameras.json';
 /** Only frames on WSDOT's own image host are registered; the catalog also
@@ -317,6 +317,10 @@ export const WSDOT_CAMERAS_URL =
  * whose own terms apply. */
 export const WSDOT_IMAGE_ORIGIN = 'https://images.wsdot.wa.gov/';
 export const DEFAULT_WSDOT_MAX_SOURCES = 250;
+/** Hard ceiling on the WSDOT catalog body. The live layer is ~400 KB; this
+ * only exists so an upstream that streams an unbounded body cannot be
+ * buffered without limit. */
+export const WSDOT_MAX_CATALOG_BYTES = 4 * 1024 * 1024;
 /** Prioritization anchors: downtown Seattle and Spokane. */
 export const WSDOT_ANCHORS = [
   { lat: 47.6062, lon: -122.3321 }, // Seattle