Changes

PSG computes trajectories with atmospheric drag now

- Constants are currently fixed to a delta3 model and haven't
  been pushed up into the API
- Cd is fixed to 0.5
- Aref is fixed to 4*pi m^3

Test on one of my rockets cut the dv discrepancy down to 100m/s
between the pad prediction and actually reaching orbit.

Signed-off-by: Lamont Granquist <lamont@scriptkiddie.org>
 (commit: 184bbdf)