fix(infrastructure): test ground below 0 m and unknown terrain correctly

The ellipsoidal horizon test hid ground below the ellipsoid (low-geoid
coasts, the Dead Sea) from a camera near it, so it is now tested at 0 m.
With the globe hidden the ground is tested as the vertical from 0 to
9,000 m against the frustum planes, in place of a 4.5 km sphere that let
in much unseen ground.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

docs/CURRENT-STATE.md

@@ -1959,11 +1959,12 @@ during continuous motion. Disable/re-enable resets selection cleanly.
 Close-up stem sizing uses camera-to-feature distance without a 5 km minimum.
 Unchanged overlay entries do not request another frame; moved positions,
 changed membership and re-enabling still publish. Ground sampling waits for
-visible globe tiles to settle, samples only markers whose ground is in the
-camera's view and above the horizon (unseen ground reads through the Earth
-and would retry every 2 s; until it is seen, a stem keeps its unsampled
-height, which can show for a stem whose base is just off screen), and uses
-valid loaded terrain as a floor while
+visible globe tiles to settle, and skips markers whose ground the camera
+cannot see (unseen ground reads through the Earth and would retry every 2 s).
+Ground is tested at loaded globe terrain, or across 0-9,000 m when the globe
+is hidden, so photoreal views skip only ground well out of view. Until it is
+seen, a stem keeps its unsampled height, which can show for a stem whose base
+is just off screen. Sampling uses valid loaded terrain as a floor while
 retaining rooftop and below-sea-level elevations. A hidden globe does not
 constrain photoreal geometry, and retry work remains bounded. Already sampled
 nearby markers follow higher settled terrain on existing bounded walks, without