Changes

ReturnToMoon: the quite stable solution

The branch in the Lambert solver was creating discontinuous "tears" in
the optimization problem.  Avoiding that branch greatly stabilizes the
problem.  I'm still not entirely certain what I just did though.

Slight tweak to the outer loop as well.

The initial guess generator is still pretty crazy, but seems to work, I
haven't come up with anything simpler.  Probably needs a better global
optimization strategy.

Signed-off-by: Lamont Granquist <lamont@scriptkiddie.org>
 (commit: 201c3d0)