Test `across()` better

[mgirlich]

Adapt big test set from dplyr

• across() should not compute on grouping columns (though this might change in dplyr: Still cannot mutate a grouping column if using across() dplyr#6127)

library(dplyr, warn.conflicts = FALSE)
library(dbplyr, warn.conflicts = FALSE)
memdb_frame(g = 1, x = 1, y = 1) %>% 
  group_by(g) %>% 
  summarise(across(everything(), ~ 0)) %>% 
  show_query()
#> <SQL>
#> SELECT `g`, 0.0 AS `g`, 0.0 AS `x`, 0.0 AS `y`
#> FROM `dbplyr_001`
#> GROUP BY `g`

^{Created on 2022-01-19 by the reprex package (v2.0.1)}

• across() should default to everything()

library(dplyr, warn.conflicts = FALSE)
suppressMessages(devtools::load_all("~/GitHub/dbplyr/"))

tibble(g = 1, x = 1, y = 1) %>% group_by(x) %>% summarise(across())
#> # A tibble: 1 × 3
#>       x     g     y
#>   <dbl> <dbl> <dbl>
#> 1     1     1     1
memdb_frame(g = 1, x = 1, y = 1) %>% group_by(x) %>% summarise(across())
#> # Source:   lazy query [?? x 1]
#> # Database: sqlite 3.37.0 [:memory:]
#>       x
#>   <dbl>
#> 1     1

^{Created on 2022-01-19 by the reprex package (v2.0.1)}