[claude] bench: measure FlatBuffers verification cost for Layout and Array

[joseph-isaacs]

Adds a divan benchmark in vortex-layout/benches/flatbuffer_verify.rs
that compares root:: (checked), root_with_opts::, and
root_unchecked:: for representative Layout and Array shapes.

Results on this machine (medians, release):

Layout (per file open):
1 x 8 (2.0 KB) -> 726 ns checked, 2.6 ns unchecked
16 x 32 (34 KB) -> 33.3 us checked
128 x 32 (295 KB) -> 277 us checked
1024 x 32 (2.5 MB)-> 2.23 ms checked

Array (per SerializedArray decode and per buffer_lengths() call):
8 fields (2.0 KB) -> 636 ns checked
100 fields(6.0 KB) -> 5.8 us checked
1000 fields(44 KB) -> 56 us checked

Key findings:

• root_unchecked is ~3 ns regardless of size; the verifier IS the cost.
• root_with_opts is the same cost as root - the Vortex VerifierOptions
  knob is a DoS bound, not a perf knob.
• Verifier walks roughly O(table count + buffer size), ~100 ns/KB.

Motivates dropping the redundant root:: re-verification inside
SerializedArray::buffer_lengths() (vortex-array/src/serde.rs:514), which
runs on an already-validated buffer.

Signed-off-by: Claude noreply@anthropic.com

[codspeed-hq]

Merging this PR will not alter performance

⚠️ 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.

⚡ 1 improved benchmark
❌ 1 regressed benchmark
✅ 1235 untouched benchmarks
🆕 10 new benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
🆕	Simulation	buffer_lengths_fixed[1000]	N/A	16.2 µs	N/A
🆕	Simulation	buffer_lengths_fixed[1]	N/A	1.7 µs	N/A
🆕	Simulation	buffer_lengths_fixed[100]	N/A	3 µs	N/A
🆕	Simulation	buffer_lengths_fixed[32]	N/A	2.1 µs	N/A
🆕	Simulation	buffer_lengths_fixed[8]	N/A	1.8 µs	N/A
🆕	Simulation	buffer_lengths_legacy_root[1000]	N/A	336.3 µs	N/A
🆕	Simulation	buffer_lengths_legacy_root[100]	N/A	39 µs	N/A
🆕	Simulation	buffer_lengths_legacy_root[1]	N/A	6.4 µs	N/A
🆕	Simulation	buffer_lengths_legacy_root[32]	N/A	16.6 µs	N/A
🆕	Simulation	buffer_lengths_legacy_root[8]	N/A	8.8 µs	N/A
❌	Simulation	chunked_varbinview_canonical_into[(100, 100)]	273.3 µs	308 µs	-11.28%
⚡	Simulation	chunked_varbinview_opt_canonical_into[(1000, 10)]	224.8 µs	187.7 µs	+19.75%

Tip

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

---

_{Comparing claude/flatbuffers-memory-safety-XKbWQ (172e0c7) with develop (7b47788)}