covariates with NAs will crash baggr with no error msg

[wwiecek]

e.g.

library(dplyr)
sch2 <- schools
sch2$cov <- c(rnorm(7), NA)
bg1 <- baggr(sch2, cov = "cov")

Provide informative error message that warns user that some values are NA and the model therefore won't run. Check for both categorical and continuous covariates and write some unit tests for this