perf: Optimize array_concat using MutableArrayData

[neilconway]

Which issue does this PR close?

• Closes Optimize array_concat to use MutableArrayData #20619 .

Rationale for this change

The current implementation of array_concat creates an ArrayRef for each row, uses Arrow's concat kernel to merge the elements together, and then uses concat again to produce the final results. This does a lot of unnecessary allocation and copying.

Instead, we can use MutableArrayData::extend to copy element ranges in bulk, which avoids much of this intermediate copying and allocation. This approach is 5-15x faster on a microbenchmark.

What changes are included in this PR?

• Add benchmark
• Improve SLT test coverage for array_concat
• Implement optimization

Are these changes tested?

Yes, and benchmarked.

Are there any user-facing changes?

No.

[neilconway]

Benchmarks:

  ┌────────────────┬────────────┬─────────────────┬─────────┐
  │   Benchmark    │ Old (main) │ New (optimized) │ Speedup │
  ├────────────────┼────────────┼─────────────────┼─────────┤
  │ 2_arrays/100   │ 20.6 µs    │ 1.73 µs         │ 11.9x   │
  ├────────────────┼────────────┼─────────────────┼─────────┤
  │ 2_arrays/1000  │ 200 µs     │ 13.1 µs         │ 15.3x   │
  ├────────────────┼────────────┼─────────────────┼─────────┤
  │ 2_arrays/10000 │ 2.02 ms    │ 126 µs          │ 16.0x   │
  ├────────────────┼────────────┼─────────────────┼─────────┤
  │ elem_size/5    │ 202 µs     │ 14.5 µs         │ 13.9x   │
  ├────────────────┼────────────┼─────────────────┼─────────┤
  │ elem_size/50   │ 224 µs     │ 18.4 µs         │ 12.2x   │
  ├────────────────┼────────────┼─────────────────┼─────────┤
  │ elem_size/500  │ 475 µs     │ 95.5 µs         │ 5.0x    │
  └────────────────┴────────────┴─────────────────┴─────────┘