Implement some() and every() helpers

[hadley]

These are more efficient than any/all plus vapply() because they can early exit. Benchmarking confirms that they are faster even in the worst case scenario (TRUE/FALSE at last entry).

x <- as.list(1:10)
bench::mark(
  # some() early-exits on first TRUE; any(vlapply) computes all
  some_early    = some(x, \(v) v > 0),
  any_vlapply   = any(vlapply(x, \(v) v > 0)),
  # worst case: no early exit possible
  every_full    = every(x, \(v) v > 0),
  all_vlapply   = all(vlapply(x, \(v) v > 0)),
  check = FALSE
)[1:4]

#>   expression       min   median `itr/sec`
#>   <bch:expr>  <bch:tm> <bch:tm>     <dbl>
#> 1 some_early  410.01ns 574.04ns  1157952.
#> 2 any_vlapply   2.38µs   2.91µs   313534.
#> 3 every_full    1.31µs   1.64µs   563264.
#> 4 all_vlapply   2.34µs   2.91µs   330469.