Merge pull request #5786 from RosettaCommons/vmullig/fix_rama_prepro_edge_case
Fix an edge case for bad input files for the rama_prepro scoreterm
Andy Watkins found a bug that produces very hard-to-diagnose errors in scoring: if a rama_prepro map has 0 for any probability entry, it creates NaNs that appear far downstream. This PR adds an error message if a rama_prepro map has 0 for any probability entry. It also switches a few more instances of `new` to use `utility::pointer::make_shared`.