Avoid splicing eagerly in bench::mark()

[lionel-]

Splicing too early with enexprs() transforms the spliced input into a pairlist, which has performance implications when interfacing with list2():

x <- as.list(1:1e5)
foo <- function(x) list2(!!!x)

bench::mark(
  lang = list2(!!!x),
  value = foo(x)
)[1:8]
#> # A tibble: 2 x 8
#>   expression      min   median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl> <int> <dbl>
#> 1 lang         41.8ms  44.73ms      21.7     781KB     2.17    10     1
#> 2 value        1.58ms   1.79ms     560.      781KB     8.74   256     4

The first commit also fixes a deprecation warning from as_tibble().

Should we also fix it in workout()? It looks like enquo() was used from the beginning so I'm not sure how much of a breaking change this would be.

[jimhester]

Yeah I think we should fix it in workout as well, I think any splicing there is likely to be unintended, as romain ran into.

[lionel-]

Now fixed in workout() as well.

[jimhester]

Thanks!