Book 1.9.1 Suggestion: Clarify reason for vector rejection

[NickOveracker]

I think there is a risk that many readers will give up trying to understand the reason why vectors that extend beyond the sphere's surface are rejected, and will proceed with a grain of uncertainty and doubt throughout the rest of the book.

The reason, to be clear, is that accepting samples that extend beyond the surface results in a sampling bias towards the corners of the cube.

There is also one other question that might be worth addressing here: why use a unit cube at all instead of sampling angles in spherical coordinates? I think that the reason for this is to avoid computing two trigonometric functions on each sample, but I am not sure.