When I removed the cube based stuff from eathdistance I accidentally left
a reference to cube in a comment in the regression test (that also
shows up in the expected output). This doesn't cause any real problem,
but people who read the comment might be confused. Attached is a diff
to remove the reference.
