feat(scalar): push byte_length validity down to its input

[joseph-isaacs]

byte_length is null-preserving: the result is null exactly when the
input is null. Previously, ByteLength did not implement the
ScalarFnVTable::validity rule, so requesting the validity of a
byte_length array fell back to is_not_null(byte_length(input)). That
fallback forces the whole function to be evaluated just to extract the
validity mask, which for compressed encodings like FSST means
decompressing the codes into a canonical VarBinView.

Implement ByteLength::validity to return the input's validity directly.
For FSST this reads the validity buffer straight from the array via
is_not_null without touching the compressed codes.

Signed-off-by: Joe Isaacs joe.isaacs@live.co.uk

[codspeed-hq]

Merging this PR will improve performance by 27.23%

⚠️ 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.

⚡ 189 improved benchmarks
❌ 3 regressed benchmarks
✅ 1334 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	bitwise_not_vortex_buffer_mut[128]	216.9 ns	275.3 ns	-21.19%
❌	Simulation	bitwise_not_vortex_buffer_mut[1024]	278.6 ns	336.9 ns	-17.31%
❌	Simulation	bitwise_not_vortex_buffer_mut[2048]	342.2 ns	400.6 ns	-14.56%
⚡	Simulation	compare[63]	360.5 µs	244.8 µs	+47.29%
⚡	Simulation	chunked_bool_canonical_into[(1000, 10)]	46.4 µs	31.7 µs	+46.04%
⚡	Simulation	compare[56]	327.6 µs	224.8 µs	+45.69%
⚡	Simulation	compare[62]	363.9 µs	250.1 µs	+45.51%
⚡	Simulation	compare[63]	371.2 µs	255.6 µs	+45.26%
⚡	Simulation	compare[60]	353.8 µs	243.7 µs	+45.16%
⚡	Simulation	compare[56]	332.4 µs	229.6 µs	+44.79%
⚡	Simulation	compare[62]	368.7 µs	254.8 µs	+44.71%
⚡	Simulation	compare[61]	362.7 µs	250.8 µs	+44.61%
⚡	Simulation	compare[60]	358.6 µs	248.5 µs	+44.31%
⚡	Simulation	compare[58]	347.3 µs	241 µs	+44.15%
⚡	Simulation	compare[59]	354.5 µs	246.3 µs	+43.97%
⚡	Simulation	compare[61]	367.5 µs	255.5 µs	+43.84%
⚡	Simulation	compare[58]	352.2 µs	245.7 µs	+43.33%
⚡	Simulation	compare[57]	345.9 µs	241.5 µs	+43.25%
⚡	Simulation	compare[59]	359.3 µs	251 µs	+43.15%
⚡	Simulation	compare[54]	330.4 µs	231.5 µs	+42.71%
...	...	...	...	...	...

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 claude/scalar-validity-optimization-k3lxyr (9e799eb) with develop (47d2041)}