In recent runs of the stack through multiBandDriver.py, the persisted forced src tables for the coadds are not getting ra and dec set properly (all entries for the coord_ra and coord_dec columns are NaN). Looking back at a run in mid-Nov, 2016, these numbers were indeed set properly in the forced tables. Assuming this was not intentional, track down the cause and fix it such that these values get set properly for the persisted forced src tables.