Return from Moon bugfixes/rewrite

[lamont-granquist]

This is mostly fixed now for the basic algorithm it could still use:

• time selector for the actual Maneuver with a dt parameter pushed into the problem solver
• hyperbolic initial orbit support for fine-tuning or changing your mind after the fact
• maybe the "N orbits in the future" support for finding a better return window
• look at uncommenting all the "hard" tests which seem to require that N-orbits-in-the-future support