Merge pull request #770 from RosettaCommons/dimaio/envdep_elec
Fixing two issues with lk_ball:
1) remove derivative discontinuity in between water positions by replacing a min with a soft min
2) implement proper cartesian derivatives for movement of the stub atoms used to build waters
The changes from result in very small changes to scoring, and no noticeable runtime changes.
Three parameters controlling the potential are exposed through the options system/
These fixes are also implemented for using lk_ball with -nblist_autoupdate.