Fix spherical unit vectors documentation

[bobmyhill]

The documentation about the use of spherical unit vectors to define boundary velocities here is incorrect and unclear.

We should say something like:
Specify velocity in a local reference frame instead of global x, y, and z. Positive velocities point up, east, and south (in 3d) or out and clockwise (in 2d). This setting only makes sense for spherical geometries.

Issue raised here.