Make primitives/entities not cull when below ellipsoid but above terrain

[IanLilleyT]

The current system culls entities/primitives when they are below the minimum radius of the earth, ~6356752 meters. In practice there are not many points on earth that are below this, except maybe https://en.wikipedia.org/wiki/Bentley_Subglacial_Trench which cesium world terrain doesn't accurately map. But with terrain exaggeration enabled or some other configuration, this becomes a problem. The proposed fix is to keep track of the minimum rendered terrain tile height and offset the scene's occluder radius by it. This information could also assist in choosing a better distance for the depth plane: #8203.

In this Sandcastle, left click to spawn red dots. In master, the dots do not appear. In this PR, they do.

This is a draft PR because there are a few unknowns:

• Do we want to include 3D tiles when calculating this minimum distance?
• Instead of limiting the offset to 0, what if it could offset upwards when all terrain heights are above 0. Any hidden downsides to this?
• What should happen to entities if the globe is missing tiles?

[cesium-concierge]

Thanks for the pull request @IanLilleyT!

• ✔️ Signed CLA found.
• ❔ CHANGES.md was not updated.
  • If this change updates the public API in any way, please add a bullet point to CHANGES.md.
• ❔ Unit tests were not updated.
  • Make sure you've updated tests to reflect your changes, added tests for any new code, and ran the code coverage tool.

Reviewers, don't forget to make sure that:

• Cesium Viewer works.
• Works in 2D/CV.
• Works (or fails gracefully) in IE11.

[lilleyse]

This could be less conservative by setting the occluder radius to the min distance rather than the minor axis minus the min height. You could have a case where an entity at the equator is 11000 meters below the ellipsoid, in which case it would create a smaller occluder than needed since the entity itself would not be below the minor axis.

In the end though we're talking about ~11000 meter difference in an already very conservative occluder (the occluder radius is the minor axis which is approx. 20000 meters smaller than the semi-major axis). If we were truly concerned about perfect occlusion tests we would use an ellipsoid vs. sphere test, though I think the math gets fuzzier and slower.