Update value of a linear predictor as function of a single coordinate change
Source:R/glm_utils.R
      update_linear_predictor.RdFunction for updating the linear predictor with n actions rather than nn_vars actions if naively doing a matrix-vector product of X %% beta
Arguments
- new_beta_j
 a
numericnew value of the j'th component of the beta parameter vector- current_beta_j
 the
numericcurrent value of the j'th component of the beta parameter vector- current_eta
 the current value of the linear predictor corresponding to the
current_betavalue- X_j
 the j'th column of the design matrix