factor_to_indicators errors with unhelpful message if a factor has only one level

[benkeser]

grbg <- data.table(y = 1:10, x = as.factor(rep("F", 10)), z = 1:10)

grbg_task <- sl3_Task$new(data = grbg,
                          covariates = c("x", "z"),
                          outcome = "y")

grbg_task$X

Obviously this is sort of a user error, but the error message is

Error in ind_ref_mat[as.numeric(x), , drop = FALSE] : 
  incorrect number of dimensions

which is not helpful. And I could see users processing data without fully checking that all levels of a factor are present (that's what I did 😳).