Option for light brightness to be proportional to amount of visual sphere the light occupies

[henryseg]

With this, lights would all have non-zero size, so no point lights. The brightness of a light as seen from a point p would be how much of the sphere of geodesic rays starting at p hit the light. This removes infinities coming from the inverse linear, quadratic, etc. models. Done simply, this just involves working out a different function for brightness based on distance from the light, which now also depends on the radius of the light.

[henryseg]

For a light of radius r that we are distance d away from, the fraction of our sky that is filled with the light is:

1 if d <= r. Otherwise:
In spherical space: 1 - sqrt(1 - sin(r)^2/sin(d)^2)
In euclidean space: 1 - sqrt(1 - r^2/d^2)
In hyperbolic space: 1 - sqrt(1 - sinh(r)^2/sinh(d)^2)

These will look similar to the "physical" model we have already at large distances, but should hopefully work better closeup. Maybe it would be good to try having one light be a really large "sun" some distance away?