Add split-collection benchmark for scan planning

[AdamGS]

Rationale for this change

Benchmarks SplitBy::Layout over written files (struct -> chunked -> flat per column) across column and chunk counts:

• cold: fresh reader tree per iteration, as the first scan over a file
• warm: reused reader tree
• cold_single_column: narrow field mask over a wide schema
• {cold,warm}_misaligned: per-column chunk granularities so no two columns share interior chunk boundaries

[codspeed-hq]

Merging this PR will improve performance by 12.39%

⚠️ 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.

⚡ 2 improved benchmarks
✅ 1975 untouched benchmarks
🆕 13 new benchmarks
⏩ 8 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	decompress[u64, (1000, 16)]	72.5 µs	64.5 µs	+12.44%
⚡	Simulation	search_index_in_range_chunked	6.1 ms	5.4 ms	+12.34%
🆕	Simulation	cold_misaligned[(64, 256)]	N/A	36 ms	N/A
🆕	Simulation	cold_misaligned[(8, 256)]	N/A	4.2 ms	N/A
🆕	Simulation	cold_single_column[(1, 1024)]	N/A	4.9 ms	N/A
🆕	Simulation	cold_single_column[(64, 256)]	N/A	1.7 ms	N/A
🆕	Simulation	cold_single_column[(8, 256)]	N/A	1.6 ms	N/A
🆕	Simulation	cold[(1, 1024)]	N/A	2.2 ms	N/A
🆕	Simulation	cold[(64, 256)]	N/A	32.8 ms	N/A
🆕	Simulation	cold[(8, 256)]	N/A	4.1 ms	N/A
🆕	Simulation	warm_misaligned[(64, 256)]	N/A	75.8 ms	N/A
🆕	Simulation	warm_misaligned[(8, 256)]	N/A	8.7 ms	N/A
🆕	Simulation	warm[(1, 1024)]	N/A	1.5 ms	N/A
🆕	Simulation	warm[(64, 256)]	N/A	71.3 ms	N/A
🆕	Simulation	warm[(8, 256)]	N/A	8.3 ms	N/A

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing adamg/split-bench (7274546) with develop (d443eab)}

Footnotes

1. 8 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. ↩