work with similar variable names more easily

[adw96]

From Gary by email:

Caught an error in formatting the regression results! Some common causes for this error message:
      (i) name conflicts between covariates and one of the categories in the data (consider renaming your covariates);
      (ii) interactions between categorical variables with specified reference levels (consider changing the reference levels outside of the regress call).

"I have variables named height, heightsp55 and heightsp65 and I put U(~height+heightsp55+heightsp65) and I got the above error, but if I rename heightsp55 and heightsp65 as hsp55 and hsp65, then U(~height+hsp55+hsp65) does not give me any error."

Could rigr just deal with this by default?