fix: emit [0] offsets for empty structural list streams (alternative to #10533)

[danking]

Summary

Alternative to #10533, based on the same develop commit (18aebea), for comparing the two empty-list representations. Fixes the same ListLayoutStrategy needs at least one chunk error for empty/null nested lists.

Changes

For an empty stream, emit the offsets sentinel [0] and let the normal child writers finish the empty elements and validity streams. This preserves a structural ListLayout at each nested list level instead of returning one typed empty chunked layout for the whole empty subtree as #10533 does.

Reserve the sentinel's sequence ID before the child EOFs so it can be written without blocking behind child completion. Nonempty input continues through the existing transposition path.

Includes the same four file-roundtrip regressions and nullable/nonnullable empty-stream checks as #10533, plus persisted-layout tests at depths 1, 2, and 5 that read back each empty level's offsets and assert [0].

Validation

The [0] implementation passed the same 76 selected tests as #10533. The extended run including the new depth/layout tests is in progress. Both changed files were formatted with CI's pinned nightly-2026-09-10 rustfmt; git diff --check passes.

Implemented with OpenAI Codex assistance.

[codspeed]

Merging this PR will degrade performance by 3.68%

⚠️ 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.

⚠️ 25 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• test_repeated_scan_scalar_at[int]
• test_scan_filter_to_arrow_array[int_float]
• test_repeated_scan_filter[int_float]
• test_scan_scalar_at[int]
• test_scan[int]
• test_repeated_scan_to_arrow_array[int_float]
• test_scan_filter[int_float]
• bench_compare_sliced_dict_primitive[(3333, 10000)]
• chunked_dict_primitive_canonical_into[u64, (1000, 1000, 10)]
• take_varbin
• …and 15 more

ℹ️ 40 benchmarks have no flame graph

The flamegraph could not be generated because we failed to resolve the symbols in the profiling trace. This happens when we fail to harvest information about language frames.

• test_repeated_scan_scalar_at[int]
• test_scan_filter_to_arrow_array[int_float]
• test_repeated_scan_filter[int_float]
• test_scan_scalar_at[int]
• test_repeated_scan_filter[int]
• test_repeated_scan_scalar_at[int_float]
• test_scan_filter_to_arrow_array[int]
• test_scan_filter[int]
• test_scan[int]
• test_repeated_scan_to_arrow_array[int_float]
• …and 30 more

⚠️ 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

⚡ 3 improved benchmarks
❌ 5 regressed benchmarks
✅ 2251 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌ ⚠️	Simulation	test_repeated_scan_scalar_at[int]	263.7 µs	355.8 µs	-25.87%
❌ ⚠️	Simulation	test_scan_filter_to_arrow_array[int_float]	3.5 ms	4.1 ms	-14.81%
❌ ⚠️	Simulation	test_repeated_scan_filter[int_float]	3.4 ms	4 ms	-13.44%
❌ ⚠️	Simulation	test_scan_scalar_at[int]	803.7 µs	905.7 µs	-11.26%
❌ ℹ️	Simulation	test_repeated_scan_filter[int]	2.5 ms	2.7 ms	-10.4%
⚡ ℹ️	Simulation	test_repeated_scan_scalar_at[int_float]	486.7 µs	396.8 µs	+22.68%
⚡	WallTime	bitpack_blocked_compress_avx512	5.6 µs	4.8 µs	+18.06%
⚡	WallTime	filter_neon[80]	2.9 µs	2.5 µs	+17.73%

Tip

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

---

_{Comparing danking:dk/fix-empty-list-zero-offset (2c31fee) with develop (18aebea)}

[mhk197]

I think we should close in favor of #10533. We shouldn't need to write dummy segments for correctness.

I know struct layout does this differently but not sure why @robert3005