bench(bitpacking): more representative unpack_indices benchmarks

[lwwmanning]

The benchmarks from #190 did not compare the three paths fairly. The compiler treated the bit width as a constant for some paths and removed work that real callers still do. This PR gives every path the same runtime bit width and covers all four integer types near the candidate cutoffs.

It also restores extraneous code comment removals in #190 and makes the new tests fail safely if an unpack method skips an output.

[codspeed-hq]

Merging this PR will degrade performance by 35.71%

❌ 1 regressed benchmark
✅ 157 untouched benchmarks
🆕 48 new benchmarks
⏩ 171 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	throughput_compress	1.5 ms	2.3 ms	-35.71%
🆕	Simulation	batched[1]	N/A	1.3 µs	N/A
🆕	Simulation	batched[32]	N/A	4 µs	N/A
🆕	Simulation	batched[48]	N/A	4.3 µs	N/A
🆕	Simulation	batched[8]	N/A	2.3 µs	N/A
🆕	Simulation	full_unpack_then_gather[1]	N/A	4.4 µs	N/A
🆕	Simulation	full_unpack_then_gather[32]	N/A	4.9 µs	N/A
🆕	Simulation	full_unpack_then_gather[48]	N/A	5.2 µs	N/A
🆕	Simulation	full_unpack_then_gather[8]	N/A	4.5 µs	N/A
🆕	Simulation	repeated_single[1]	N/A	1 µs	N/A
🆕	Simulation	repeated_single[32]	N/A	4 µs	N/A
🆕	Simulation	repeated_single[48]	N/A	4.5 µs	N/A
🆕	Simulation	repeated_single[8]	N/A	2.1 µs	N/A
🆕	Simulation	batched[1]	N/A	1.3 µs	N/A
🆕	Simulation	batched[64]	N/A	6.2 µs	N/A
🆕	Simulation	batched[8]	N/A	2.4 µs	N/A
🆕	Simulation	batched[80]	N/A	6.6 µs	N/A
🆕	Simulation	full_unpack_then_gather[1]	N/A	7.3 µs	N/A
🆕	Simulation	full_unpack_then_gather[64]	N/A	8.5 µs	N/A
🆕	Simulation	full_unpack_then_gather[8]	N/A	7.4 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 wm/runtime-unpack-benchmarks (d85faa0) with develop (069f29d)}

Footnotes

1. 171 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. ↩

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-01T20:54:51.158963Z	d85faa0	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.