add_p throws error if all values are NA for a column

[fh-jsnider]

In 1.4.0, if all values are missing for a given column, instead of not producing a p-value, adding a p-value with add_p() returns an unhelpful error message:

 library(tidyverse)
 mtcars %>% 
 mutate(error_var = NA) %>%
 gtsummary::tbl_summary(by = "cyl") %>% 
 gtsummary::add_p()
#> Error: Problem with `mutate()` input `test_name`.
#> x Result 11 must be a single string, not NULL of length 0
#> ℹ Input `test_name` is `map_chr(.data$test_info, ~pluck(.x, "test_name"))`.

^{Created on 2021-05-04 by the reprex package (v0.3.0)}

Previous versions of gtsummary would give a warning message around the impossibility of creating a p-value.

[ddsjoberg]

Thank you so much for the repot @fh-jsnider !

I've added a fix to the dev version. If you'd like to continue using the CRAN release, you'll just need to assign a test to the variable with all missing variables. The issue arose when the missing column was not assigned a default test.

[fh-jsnider]

Thanks so much! Appreciate the speedy work :)