Replace stats::ave()

[teunbrand]

This PR aims to fix #6696.

Essentially, stats::ave() created 0-length groups when used with a factor-grouping where some levels are absent. This then causes troubles for functions such as min() and max() that throw warnings when given 0-length data. We replace stats::ave() with the example from ?vec_chop. The vctrs-based solution is faster than stats::ave(), so I replaced all instances, not just the problematic one.

devtools::load_all("~/packages/ggplot2/")
#> ℹ Loading ggplot2
#> Warning: package 'testthat' was built under R version 4.4.2

nrow(diamonds)
#> [1] 53940
bench::mark(
  ave(diamonds$carat, diamonds$color, FUN = mean),
  vec_ave(diamonds$carat, diamonds$color, mean), 
  min_iterations = 100
)
#> # A tibble: 2 × 6
#>   expression                             min median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr>                        <bch:tm> <bch:>     <dbl> <bch:byt>    <dbl>
#> 1 ave(diamonds$carat, diamonds$col…   2.16ms 2.41ms      389.    2.31MB     30.4
#> 2 vec_ave(diamonds$carat, diamonds…  998.8µs 1.15ms      830.    2.37MB     52.6

nrow(mtcars)
#> [1] 32
bench::mark(
  ave(mtcars$mpg, mtcars$cyl, FUN = mean),
  vec_ave(mtcars$mpg, mtcars$cyl, mean),
  min_iterations = 100
)
#> # A tibble: 2 × 6
#>   expression                            min  median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr>                        <bch:t> <bch:t>     <dbl> <bch:byt>    <dbl>
#> 1 ave(mtcars$mpg, mtcars$cyl, FUN … 124.7µs 135.4µs     6997.    1.48KB     6.19
#> 2 vec_ave(mtcars$mpg, mtcars$cyl, …  29.5µs  32.5µs    28604.    1.04KB     8.58

^{Created on 2025-10-13 with reprex v2.1.1}