Right now, when the initial parameters for a given fit are initialized with the results (parameter cube) from a previous fit, the broad emission line components are not really initialized because the get_feats() function does not return their parameters (VAL, VARY, etc.), only that the components exist, so the make_feat_pars() function can recreate the lmfit parameters as if the feature parameters were provided by the init_feats() function, i.e., via tables (where double components are only specified by their existence, from the column DOUB).
Right now, when the initial parameters for a given fit are initialized with the results (parameter cube) from a previous fit, the broad emission line components are not really initialized because the get_feats() function does not return their parameters (VAL, VARY, etc.), only that the components exist, so the make_feat_pars() function can recreate the lmfit parameters as if the feature parameters were provided by the init_feats() function, i.e., via tables (where double components are only specified by their existence, from the column DOUB).