predict and residuals have odd behavior

Check out
```

dat = tibble( siteID = rep( LETTERS[1:4], each=10 ),
              X = rnorm( 40 ),
              Y = X + rnorm( 40 ) )
mod = lm_robust( Y ~ X + siteID, data=dat )
mod
predict( mod )
residuals( mod )
```

Should these return values?