across() not operating sequentially

[romainfrancois]

I added some tests too, code from #5770:

library(palmerpenguins); library(tidyverse)
(tbl1 <- penguins %>%
    group_by(species, island, sex) %>%
    summarize(n = n()) %>%
    pivot_wider(names_from = sex, values_from = n))
#> `summarise()` has grouped output by 'species', 'island'. You can override using the `.groups` argument.
#> # A tibble: 5 x 5
#> # Groups:   species, island [5]
#>   species   island    female  male  `NA`
#>   <fct>     <fct>      <int> <int> <int>
#> 1 Adelie    Biscoe        22    22    NA
#> 2 Adelie    Dream         27    28     1
#> 3 Adelie    Torgersen     24    23     5
#> 4 Chinstrap Dream         34    34    NA
#> 5 Gentoo    Biscoe        58    61     5
(tbl2 <- tbl1 %>%
    ungroup() %>%
    rowwise(species, island) %>%
    mutate(across(everything(), ~ . / sum(c_across(everything()), na.rm = T))))
#> # A tibble: 5 x 5
#> # Rowwise:  species, island
#>   species   island    female  male    `NA`
#>   <fct>     <fct>      <dbl> <dbl>   <dbl>
#> 1 Adelie    Biscoe     0.5   0.5   NA     
#> 2 Adelie    Dream      0.482 0.5    0.0179
#> 3 Adelie    Torgersen  0.462 0.442  0.0962
#> 4 Chinstrap Dream      0.5   0.5   NA     
#> 5 Gentoo    Biscoe     0.468 0.492  0.0403
(tbl3 <- tbl1 %>%
    ungroup() %>%
    rowwise(species, island) %>%
    mutate(across(female:male, ~ . / sum(c_across(female:male), na.rm = T))))
#> # A tibble: 5 x 5
#> # Rowwise:  species, island
#>   species   island    female  male  `NA`
#>   <fct>     <fct>      <dbl> <dbl> <int>
#> 1 Adelie    Biscoe     0.5   0.5      NA
#> 2 Adelie    Dream      0.491 0.509     1
#> 3 Adelie    Torgersen  0.511 0.489     5
#> 4 Chinstrap Dream      0.5   0.5      NA
#> 5 Gentoo    Biscoe     0.487 0.513     5

^{Created on 2021-02-18 by the reprex package (v0.3.0)}