perf: Optimize array_has() for scalar needle

[neilconway]

Which issue does this PR close?

• Closes Optimize array_has() for scalar needle #20377.

Rationale for this change

compare_with_eq() checks for matching array elements via a single pass across the entire flat values buffer, which is reasonably fast. The previous implementation then determined per-row results by creating a BooleanArray slice for each row and calling true_count() to check for any matches. It turns out that that's quite a lot of per-row work.

Instead, we use BooleanBuffer::set_indices() to iterate over the set bits in the comparison result in a single forward pass. We walk this iterator in lockstep with the row offsets to determine whether each row contains a match, which does much less work per-row.

This can be substantially faster, especially for short arrays. For example, for 10-element arrays of int64, it is 3-5x faster than the previous approach. 10-element string arrays are 1.6-4.8x faster. The improvement is smaller but non-zero for larger arrays (e.g., ~1.2x faster for 500 element arrays).

What changes are included in this PR?

In addition to the optimization, this commit adjusts the array_has benchmark code to actually benchmark array_has evaluation (!). The previous benchmark just constructed an Expr.

Are these changes tested?

Yes. Passes existing tests. Performance validated via several benchmark runs.

Are there any user-facing changes?

No.

[neilconway]

Benchmarks:

  group                                       vanilla                                opt
  -----                                       ----                                   ------
  array_has_all/all_found_small_needle/10     1.00      4.6±0.23ms        ? ?/sec    1.00      4.6±0.04ms        ? ?/sec
  array_has_all/all_found_small_needle/100    1.00     11.2±0.12ms        ? ?/sec    1.01     11.4±0.09ms        ? ?/sec
  array_has_all/all_found_small_needle/500    1.01     46.2±0.58ms        ? ?/sec    1.00     45.8±1.09ms        ? ?/sec
  array_has_all/not_all_found/10              1.00      4.3±0.04ms        ? ?/sec    1.00      4.3±0.05ms        ? ?/sec
  array_has_all/not_all_found/100             1.00     10.3±0.20ms        ? ?/sec    1.02     10.5±0.06ms        ? ?/sec
  array_has_all/not_all_found/500             1.01     41.4±0.49ms        ? ?/sec    1.00     41.0±0.89ms        ? ?/sec
  array_has_all_strings/all_found/10          1.07      4.0±0.07ms        ? ?/sec    1.00      3.8±0.03ms        ? ?/sec
  array_has_all_strings/all_found/100         1.00     11.7±0.21ms        ? ?/sec    1.01     11.8±0.10ms        ? ?/sec
  array_has_all_strings/all_found/500         1.02     48.5±1.75ms        ? ?/sec    1.00     47.7±2.52ms        ? ?/sec
  array_has_all_strings/not_all_found/10      1.00      2.7±0.04ms        ? ?/sec    1.02      2.8±0.04ms        ? ?/sec
  array_has_all_strings/not_all_found/100     1.03     10.5±0.26ms        ? ?/sec    1.00     10.2±0.12ms        ? ?/sec
  array_has_all_strings/not_all_found/500     1.00     57.8±0.96ms        ? ?/sec    1.00     57.6±0.81ms        ? ?/sec
  array_has_any/no_match/10                   1.07      5.4±0.13ms        ? ?/sec    1.00      5.0±0.22ms        ? ?/sec
  array_has_any/no_match/100                  1.00     17.6±0.45ms        ? ?/sec    1.02     18.1±0.21ms        ? ?/sec
  array_has_any/no_match/500                  1.00     78.4±1.43ms        ? ?/sec    1.03     80.7±0.62ms        ? ?/sec
  array_has_any/some_match/10                 1.01      4.6±0.05ms        ? ?/sec    1.00      4.5±0.09ms        ? ?/sec
  array_has_any/some_match/100                1.00     10.9±0.10ms        ? ?/sec    1.03     11.2±0.15ms        ? ?/sec
  array_has_any/some_match/500                1.10     47.9±0.64ms        ? ?/sec    1.00     43.6±0.61ms        ? ?/sec
  array_has_any_strings/no_match/10           1.00      3.6±0.05ms        ? ?/sec    1.02      3.7±0.07ms        ? ?/sec
  array_has_any_strings/no_match/100          1.00     17.5±0.22ms        ? ?/sec    1.00     17.5±0.28ms        ? ?/sec
  array_has_any_strings/no_match/500          1.03    112.5±1.99ms        ? ?/sec    1.00    109.6±1.89ms        ? ?/sec
  array_has_any_strings/some_match/10         1.00      3.3±0.04ms        ? ?/sec    1.13      3.7±0.08ms        ? ?/sec
  array_has_any_strings/some_match/100        1.00     10.4±0.16ms        ? ?/sec    1.04     10.9±0.13ms        ? ?/sec
  array_has_any_strings/some_match/500        1.00     42.6±1.31ms        ? ?/sec    1.00     42.5±1.06ms        ? ?/sec
  array_has_i64/found/10                      3.14    516.1±8.76µs        ? ?/sec    1.00    164.1±4.76µs        ? ?/sec
  array_has_i64/found/100                     1.57  1043.2±25.75µs        ? ?/sec    1.00   666.3±15.72µs        ? ?/sec
  array_has_i64/found/500                     1.19      3.7±0.05ms        ? ?/sec    1.00      3.1±0.18ms        ? ?/sec
  array_has_i64/not_found/10                  5.27    514.7±4.70µs        ? ?/sec    1.00     97.7±3.40µs        ? ?/sec
  array_has_i64/not_found/100                 1.85  1035.2±11.34µs        ? ?/sec    1.00   559.5±17.33µs        ? ?/sec
  array_has_i64/not_found/500                 1.22      3.7±0.10ms        ? ?/sec    1.00      3.0±0.09ms        ? ?/sec
  array_has_strings/found/10                  1.61   996.1±13.42µs        ? ?/sec    1.00    618.1±6.67µs        ? ?/sec
  array_has_strings/found/100                 1.18      2.5±0.03ms        ? ?/sec    1.00      2.1±0.10ms        ? ?/sec
  array_has_strings/found/500                 1.13     10.3±0.82ms        ? ?/sec    1.00      9.1±0.80ms        ? ?/sec
  array_has_strings/not_found/10              4.82   550.1±33.51µs        ? ?/sec    1.00    114.2±3.77µs        ? ?/sec
  array_has_strings/not_found/100             1.15      5.3±0.06ms        ? ?/sec    1.00      4.6±0.13ms        ? ?/sec
  array_has_strings/not_found/500             1.05     14.1±0.22ms        ? ?/sec    1.00     13.4±0.43ms        ? ?/sec