Improve sample performance for small sets

[rstub]

fixes #52

Running my first test from #52 (on a slower CPU) brings dqsample.int closer to sample.int for small sets while not compromising performance for larger ones:

library(dqrng)

m <- 1e6
n <- 1e4
prob <- dqrunif(m)
bm <- bench::mark(sample.int(m, n, replace = TRUE, prob = prob),
                  dqsample.int(m, n, replace = TRUE, prob = prob),
                  check = FALSE)
bm[, 1:4]
#> # A tibble: 2 × 4
#>   expression                                           min   median `itr/sec`
#>   <bch:expr>                                      <bch:tm> <bch:tm>     <dbl>
#> 1 sample.int(m, n, replace = TRUE, prob = prob)    13.24ms  13.95ms      70.8
#> 2 dqsample.int(m, n, replace = TRUE, prob = prob)   2.49ms   2.54ms     388.


n <- 1e4
m <- 1e1
prob <- dqrunif(m)
bm <- bench::mark(sample.int(m, n, replace = TRUE, prob = prob),
                  dqsample.int(m, n, replace = TRUE, prob = prob),
                  check = FALSE)
bm[, 1:4]
#> # A tibble: 2 × 4
#>   expression                                           min   median `itr/sec`
#>   <bch:expr>                                      <bch:tm> <bch:tm>     <dbl>
#> 1 sample.int(m, n, replace = TRUE, prob = prob)      227µs    254µs     3724.
#> 2 dqsample.int(m, n, replace = TRUE, prob = prob)    299µs    336µs     2709.

^{Created on 2023-08-18 with reprex v2.0.2}

[rstub]

This is handled in #72