Add Validity::definitely_all_invalid and use it for AllInvalid fast paths

[joseph-isaacs]

Summary

PR 3 of a 4-PR stack (stacked on #8334) preparing Validity for lazy validity arrays.

Replaces the ~18 scattered matches!(.., Validity::AllInvalid) fast-path checks with a named helper, symmetric with definitely_no_nulls() from #8333. The name makes the conservative semantics explicit: a Validity::Array may still resolve to all-invalid once executed, so false means "unknown without compute", not "definitely has valid values".

Converted sites: the 8 DuckDB exporters, CUDA kernels (fsst, runend, datetime-parts, dynamic dispatch), list_contains, fill_null kernel, primitive top_value, dict codes, masked vtable, and varbin. The three assert!(matches!(...)) sites in tests keep the raw matches!, since they intentionally assert the exact variant.

No behavior change — purely a naming/readability refactor.

Checks

• cargo nextest run -p vortex-array (2962 passed)
• cargo nextest run -p vortex-duckdb (196 passed; test_geometry/over_http excluded — they need network access unavailable in this sandbox)
• cargo clippy -p vortex-array -p vortex-duckdb --all-targets, cargo +nightly fmt --all
• Could not build/test vortex-cuda locally (CUDA toolkit unavailable in this environment).

https://claude.ai/code/session_01VPQ7dfZtijfrsjAipwXvEj

---

Generated by Claude Code

[codspeed-hq]

Merging this PR will improve performance by 15.07%

⚠️ 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.

⚡ 5 improved benchmarks
❌ 1 regressed benchmark
✅ 1520 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	cuda/bitpacked_u8/unpack/3bw[100M]	298.6 µs	349.5 µs	-14.57%
⚡	Simulation	chunked_bool_canonical_into[(1000, 10)]	47 µs	31.8 µs	+47.79%
⚡	Simulation	chunked_varbinview_canonical_into[(1000, 10)]	198.6 µs	161.8 µs	+22.73%
⚡	Simulation	chunked_varbinview_into_canonical[(1000, 10)]	213.2 µs	177.2 µs	+20.3%
⚡	Simulation	chunked_varbinview_canonical_into[(100, 100)]	308.9 µs	274.4 µs	+12.57%
⚡	Simulation	chunked_varbinview_into_canonical[(100, 100)]	362.1 µs	327.2 µs	+10.66%

Tip

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

---

_{Comparing claude/cool-bardeen-l8jlsy-3-definitely-all-invalid (97b60ac) with claude/cool-bardeen-l8jlsy-2-mask-eq (4813bec)}