another merge from my branch rhiju/dms_score, which introduces rna_chem_map_lores score term. This allows chemical mapping data to guide fragment assembly, and is being tested in the RNA puzzles blind trials. Integration tests show changes in KIC_refine, SnugDock, antibody_H3, and a few other tests that look like floating point deviations. This has happened to me before when I update ScoreType.hh -- I will pay attention to whether there is an effect on the test server...