Ignore empty constant chunks in aggregate states

[connortsui20]

Tracking Issue: #10177

Summary

A zero-length constant array still carries a scalar, but that scalar contributes no values to an aggregate. The Constant arms for MinMax, constantness, and sortedness currently observe it anyway, creating extrema or stream boundaries that can change results for later chunks.

Changes

Skip those Constant arms when the input is empty, preserving the empty partial as the merge identity. An empty input has null extrema, is not constant, and is sorted. Empty chunks also leave the first and last values of a nonempty stream unchanged.

This fixes existing accumulator behavior and can be reviewed independently of the statistics migration stack.

[codspeed]

Merging this PR will regress 9 benchmarks

⚠️ 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.

⚡ 15 improved benchmarks
❌ 9 regressed benchmarks
✅ 2081 untouched benchmarks
⏩ 503 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	filter_powerlaw_by_random[250000]	131.8 µs	169.4 µs	-22.2%
❌	Simulation	take_filter_primitive_nullable_slice_mask_random_indices[4096, 1000]	96.8 µs	122.9 µs	-21.22%
❌	Simulation	take_filter_primitive_nullable_slice_mask_random_indices[16384, 1000]	115.4 µs	145 µs	-20.43%
❌	Simulation	take_filter_list_slice_mask_sequential_indices[768, 50]	128.5 µs	150.5 µs	-14.64%
❌	Simulation	take_filter_list_slice_mask_sequential_indices[256, 50]	129.9 µs	150.8 µs	-13.83%
❌	Simulation	filter_powerlaw_by_random[10000]	30.2 µs	34.1 µs	-11.54%
❌	Simulation	filter_powerlaw_by_mostly_false[10000]	29.3 µs	33.1 µs	-11.35%
❌	Simulation	filter_random_by_mostly_false[10000]	29.4 µs	33 µs	-11.04%
❌	Simulation	filter_random_by_random[10000]	30.3 µs	34 µs	-10.91%
⚡	Simulation	density_sweep_single_slice[0.9999]	85 µs	24.1 µs	×3.5
⚡	Simulation	density_sweep_single_slice[0.999]	84.7 µs	24.1 µs	×3.5
⚡	Simulation	density_sweep_single_slice[0.95]	82.8 µs	24.1 µs	×3.4
⚡	Simulation	density_sweep_single_slice[0.99]	82.6 µs	24.1 µs	×3.4
⚡	Simulation	density_sweep_single_slice[0.5]	64.6 µs	24.2 µs	×2.7
⚡	Simulation	density_sweep_single_slice[0.9]	49.4 µs	24.2 µs	×2
⚡	Simulation	density_sweep_single_slice[0.1]	48.5 µs	24.2 µs	×2
⚡	Simulation	density_sweep_single_slice[0.05]	46.7 µs	24.2 µs	+93.04%
⚡	Simulation	density_sweep_single_slice[0.01]	41.5 µs	24.2 µs	+71.63%
⚡	Simulation	density_sweep_random[0.02]	89.3 µs	64.6 µs	+38.22%
⚡	Simulation	patterns_i128[Contiguous]	30 µs	21.9 µs	+37.02%
...	...	...	...	...	...

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/stats-empty-constant-aggregates (e5061a4) with develop (8ff4b4a)²}

Footnotes

1. 503 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 (23a59e3) during the generation of this report, so 8ff4b4a was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩