add facility for conducting main effect contrasts for 'terms' where continuous variables are crossed with factor