feat: implement .before and .after for mutate()

[Yousa-Mirage]

As the title, this PR implemented the .before and .after arguments for mutate(), so all arguments of mutate() are consistent with dplyr now.

During the implementation, I also found a bug of relocate() because .before and .after are based on it. See:

library(dplyr)
library(tidypolars)

df <- tibble::tibble(
  a = 1,
  petal_1 = 2,
  petal_2 = 3,
  z = 4
)
df_pl <- as_polars_df(df)

df |> relocate(z, .before = starts_with("petal"))
#> # A tibble: 1 × 4
#>       a     z petal_1 petal_2
#>   <dbl> <dbl>   <dbl>   <dbl>
#> 1     1     4       2       3
df_pl |> relocate(z, .before = starts_with("petal"))
#> Warning in seq_len(limit): first element used of 'length.out' argument
#> Error in `seq_len()`:
#> ! argument must be coercible to non-negative integer

df |> relocate(z, .before = starts_with("not_found"))
#> # A tibble: 1 × 4
#>       z     a petal_1 petal_2
#>   <dbl> <dbl>   <dbl>   <dbl>
#> 1     4     1       2       3
df_pl |> relocate(z, .before = starts_with("not_found"))
#> Warning in seq_len(limit): first element used of 'length.out' argument
#> Error in `seq_len()`:
#> ! argument must be coercible to non-negative integer

I fixed this bug as well, just change relocate.R from

    limit <- which(names_data == .before) - 1

to

    limit <- if (is.null(.before)) {
      0
    } else {
      which(names_data == .before[1]) - 1
    }

[etiennebacher]

Thanks, LGTM, there are just minor comments, merge conflicts to fix, and snapshots to update.