fix: keep clamped traffic bounds monotonic across the antimeridian

The previous fix wrapped the span correctly but normalized the
returned west/east back into [-180, 180), which re-inverts the box
whenever it crosses the antimeridian (west > east again). Downstream
consumers (getBoundsCenter, boundsOverlap) assume west <= east and
silently misbehave on an inverted box: the fetch center lands on the
opposite side of the planet, and the overlap gate never skips a
re-fetch.

Drop the normalization and leave west/east unnormalized instead, so
the box stays monotonic. The only remaining edge case (coordinates
outside the conventional range reaching the Overpass query builder)
is a visible, separate problem rather than a silently wrong center.

Addresses review feedback on #392.

src/data/trafficBounds.js

@@ -129,17 +129,6 @@ function longitudeSpanDeg(west, east) {
   return raw >= 0 ? raw : raw + 360;
 }
 
-/**
- * Normalize a longitude into [-180, 180). Matches the wrap idiom already
- * used by `destinationPoint()` above.
- *
- * @param {number} lon - Longitude in degrees.
- * @returns {number} Longitude normalized to [-180, 180).
- */
-function normalizeLonDeg(lon) {
-  return ((lon + 540) % 360) - 180;
-}
-
 /**
  * Clamp a bounding box's spans to `maxSpanDeg` and recenter it on `center`.
  *
@@ -147,9 +136,12 @@ function normalizeLonDeg(lon) {
  * but centers the box on the derived look-at point instead of the view
  * rectangle's midpoint. Idempotent when `center` is the box's own midpoint.
  *
- * Longitude is handled as a cyclic axis: the input span wraps correctly
- * across the antimeridian, and the output west/east are normalized back into
- * [-180, 180) instead of drifting outside the normal coordinate range.
+ * Longitude is handled as a cyclic axis for the span calculation, so the
+ * input span wraps correctly across the antimeridian. The output west/east
+ * are deliberately left unnormalized (may fall outside [-180, 180)) so that
+ * `west <= east` always holds — callers such as `getBoundsCenter` and
+ * `boundsOverlap` rely on that invariant and would silently misbehave on a
+ * normalized-but-inverted box (see #392 review).
  *
  * @param {{south:number, west:number, north:number, east:number}} bounds
  *   Source bounds (span donor).
@@ -163,7 +155,7 @@ export function clampBoundsAroundCenter(bounds, center, maxSpanDeg = 0.05) {
   return {
     south: center.lat - latSpan / 2,
     north: center.lat + latSpan / 2,
-    west: normalizeLonDeg(center.lon - lonSpan / 2),
-    east: normalizeLonDeg(center.lon + lonSpan / 2),
+    west: center.lon - lonSpan / 2,
+    east: center.lon + lonSpan / 2,
   };
 }

src/data/trafficBounds.test.mjs

@@ -124,14 +124,15 @@ test('span clamp handles bounds crossing the antimeridian', () => {
 
   const result = clampBoundsAroundCenter(bounds, center, 0.05);
 
-  assert.ok(result.west >= -180 && result.west < 180);
-  assert.ok(result.east >= -180 && result.east < 180);
+  // west/east are deliberately left unnormalized (may exceed 180) so that
+  // west <= east always holds for downstream consumers (getBoundsCenter,
+  // boundsOverlap) that assume a monotonic box. See #392 review.
+  assert.ok(result.west <= result.east, `west/east not monotonic: ${result.west}, ${result.east}`);
+  assert.ok(Math.abs(result.east - result.west - 0.04) < 1e-12);
 
-  // Wrapped longitude span should remain about 0.04 degrees.
-  const span =
-    result.east >= result.west
-      ? result.east - result.west
-      : result.east - result.west + 360;
-
-  assert.ok(Math.abs(span - 0.04) < 1e-12);
+  // The plain-mean center a caller like getBoundsCenter computes must land
+  // back on the antimeridian near the requested center, not ~180 degrees
+  // away on the opposite side of the planet.
+  const center2 = (result.west + result.east) / 2;
+  assert.ok(Math.abs(center2 - center.lon) < 1e-9, `center drifted: ${center2}`);
 });