bench(array): measure RowFn Boolean retry in CI

[connortsui20]

Summary

Adds a deferred Boolean RowFn benchmark to establish the CI baseline for #9986. It uses the multiversioned collector over partially valid input and covers four cases: an accepted dense attempt and a null-only retry, each with two columns and with a constant left operand against a column.

Changes

Uses the existing native AVX2, AVX-512, and NEON jobs, so there are 12 benchmarks in total. Each timed iteration executes eight 16,384-row batches to clear the wall-time floor. Fixture and execution-context setup and result destruction stay outside timing. Backtraces follow the CI setting. This layer changes no production code.

An earlier revision ran 72 cases (both collector flags, constants in either position, all-valid input, and observable failures). All of them completed in native CI, with medians from 32.31 to 490.6 µs and a maximum printed sample of 712.1 µs.

Comparison with #9986 and raw results come from that earlier revision.

[codspeed]

Merging this PR will regress 1 benchmark

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ 4 benchmarks measured no execution time

Nothing ran under measurement, usually because the compiler removed the code under test. These results are not comparable, so they count as unchanged.

Preventing compiler optimizations

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 4 improved benchmarks
❌ 1 regressed benchmark
✅ 2203 untouched benchmarks
🆕 24 new benchmarks
⏩ 329 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	scalar_subtract_neon	12.3 µs	13.7 µs	-10.14%
⚡	Simulation	take_fsl_nullable_random[16, 100]	97.1 µs	50.4 µs	+92.78%
⚡	Simulation	take_fsl_f16_random[256, 100]	141.7 µs	85.7 µs	+65.33%
⚡	Simulation	take_chunked_fsl_sorted[32, 64]	252.9 µs	196.6 µs	+28.62%
⚡	Simulation	take_fsl_random[64, 100]	144.5 µs	125.4 µs	+15.2%
🆕	WallTime	decode_avx2[8192, (External, AllValid)]	N/A	166.2 µs	N/A
🆕	WallTime	decode_avx2[8192, (External, OneNullInEight)]	N/A	147.1 µs	N/A
🆕	WallTime	decode_avx2[8192, (Inline, AllValid)]	N/A	102.9 µs	N/A
🆕	WallTime	decode_avx2[8192, (Inline, OneNullInEight)]	N/A	95.1 µs	N/A
🆕	WallTime	deferred_bool_avx2[16384, (Columns, NullOnlyFailure)]	N/A	344.8 µs	N/A
🆕	WallTime	deferred_bool_avx2[16384, (Columns, PartialAccepted)]	N/A	110.2 µs	N/A
🆕	WallTime	deferred_bool_avx2[16384, (ConstantLhs, NullOnlyFailure)]	N/A	454.6 µs	N/A
🆕	WallTime	deferred_bool_avx2[16384, (ConstantLhs, PartialAccepted)]	N/A	206.3 µs	N/A
🆕	WallTime	decode_avx512[8192, (External, AllValid)]	N/A	171.1 µs	N/A
🆕	WallTime	decode_avx512[8192, (External, OneNullInEight)]	N/A	153.3 µs	N/A
🆕	WallTime	decode_avx512[8192, (Inline, AllValid)]	N/A	99.7 µs	N/A
🆕	WallTime	decode_avx512[8192, (Inline, OneNullInEight)]	N/A	89.8 µs	N/A
🆕	WallTime	deferred_bool_avx512[16384, (Columns, NullOnlyFailure)]	N/A	275.3 µs	N/A
🆕	WallTime	deferred_bool_avx512[16384, (Columns, PartialAccepted)]	N/A	42.3 µs	N/A
🆕	WallTime	deferred_bool_avx512[16384, (ConstantLhs, NullOnlyFailure)]	N/A	469.5 µs	N/A
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ct/row-fn-bool-bench (81ace30) with develop (36cb75d)²}

Footnotes

1. 329 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

2. No successful run was found on develop (1eb5b43) during the generation of this report, so 36cb75d was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩