Scaled values in fmt_scientific() can result in x 10^0-type output

[rich-iannone]

The fmt_scientific() formatter is designed to internally label numeric values that would lead to power-of-zero figures (e.g., 1.53 x 10^0 should just be 1.53). This results in the exclusion of those labeled numbers in getting a x 10^y part (which would end up being x 10^0).

However, the current method can result in incorrect formatting when values are scaled with scale_by != 1. Some scaled numbers incorrectly get formatted with a x 10^0 and others don't get a required x 10^y part. Here is an example:

dplyr::tibble(
  unscaled_num = c(0.1, 0.5, 1.5, 7.5, 10.5),
  unscaled_sci = c(0.1, 0.5, 1.5, 7.5, 10.5),
  scaled_num = c(0.1, 0.5, 1.5, 7.5, 10.5),
  scaled_sci = c(0.1, 0.5, 1.5, 7.5, 10.5)
) %>%
  gt() %>%
  fmt_number(
    columns = vars(unscaled_num)
  ) %>%
  fmt_scientific(
    columns = vars(unscaled_sci),
  ) %>%
  fmt_number(
    columns = vars(scaled_num),
    scale_by = 3
  ) %>%
  fmt_scientific(
    columns = vars(scaled_sci),
    scale_by = 3
  )

where rows 2 and 4 are incorrectly formatted. The solution is to scale the incoming values first, and then perform the checks for excluding the x 10^0 part.