Possible nest() bug when acting on grouped_df object and passing nesting columns

nest_grouped_df.R

reprex::reprex({
  library(magrittr)
  library(tibble)
  suppressMessages(library(tidyr))
  suppressMessages(library(dplyr))

  # Option 1
  # Proper nesting of cyl using group_by
  df1 <- mtcars %>%
    group_by(cyl) %>%
    nest()
  df1
  names(df1$data[[1]])

  # Option 2
  # Proper nesting of cyl using nest alone
  df2 <- mtcars %>%
    as_tibble() %>%
    nest(-cyl)
  df2
  names(df2$data[[1]])

  # Option 3
  # Both. Assume you used dplyr::group_by(), forgot to
  # ungroup and then wanted to tidyr::nest() on `gear`
  df3 <- mtcars %>%
    as_tibble() %>%
    group_by(cyl) %>%
    nest(-gear)
  df3
  names(df3$data[[1]])
  # In this case the group_by supercedes the nest argument, which is ok
  # The problem is `gear` is dropped from the nested data frame
  # yet gear is not part of the nesting

  # Expected (?) Behavior
  mtcars %>%
    as_tibble() %>%
    tidyr::nest(-cyl, -gear)
})