Fix nested scalar hashing

[connortsui20]

Fixes #8744

Scalar equality ignores dtype nullability recursively, but hashing only removed top-level nullability. The typed scalar views had the same mismatch, so equal nested values could produce different hashes.

• adds recursive nullability-insensitive dtype hashing matching DType::eq_ignore_nullability
• uses it for Scalar, ListScalar, StructScalar, and ExtScalar
• covers nested list, struct, Union dtype, and extension storage nullability regressions

This is kind of ugly but I don't think there is a better way to do this

[codecov]

Codecov Report

❌ Patch coverage is 95.65217% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.82%. Comparing base (17f843a) to head (76745ac).
⚠️ Report is 1 commits behind head on develop.

Files with missing lines	Patch %	Lines
vortex-array/src/dtype/dtype_impl.rs	87.50%	6 Missing ⚠️

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[github-actions]

Polar Signals Profiling Results

Latest Run

Status	Commit	Job	Attempt	Link
🟢 Done	277b386		1	Explore Profiling Data

Previous Runs (1)

Status	Commit	Job	Attempt	Link
🟢 Done	76745ac		1	Explore Profiling Data

---

Powered by Polar Signals Cloud

[github-actions]

Benchmarks: Vortex queries

Verdict: No clear signal (low confidence)
Attributed Vortex impact: -5.8%
Engines: DataFusion No clear signal (-7.8%, medium confidence) · DuckDB No clear signal (-3.8%, medium confidence)
Vortex (geomean): 0.925x ➖
Parquet (geomean): 0.953x ➖
Shifts: Parquet (control) -4.7% · Median polish -6.5%

How to read Verdict and Engines

• Verdict: Overall PR-level signal after subtracting baseline drift estimated from Parquet control rows. It can be Likely improvement, Likely regression, or No clear signal.
• Engines: Per-engine attribution. DataFusion is compared against DataFusion/Parquet controls; DuckDB is compared against DuckDB/Parquet controls. This answers whether each engine improved or regressed independently.
• Confidence: Based on directional consistency, share of rows above the noise floor, and control-run noise.

---

datafusion / vortex-file-compressed (0.869x ✅, 2↑ 0↓)

name	PR 277b386 (ns)	base 928738e (ns)	ratio (PR/base)
vortex_q00/datafusion:vortex-file-compressed 🚀	9447509	10725993	0.88
vortex_q01/datafusion:vortex-file-compressed 🚀	6300238	7350345	0.86

datafusion / parquet (0.943x ➖, 0↑ 0↓)

name	PR 277b386 (ns)	base 928738e (ns)	ratio (PR/base)
vortex_q00/datafusion:parquet	19917422	21355066	0.93
vortex_q01/datafusion:parquet	4556845	4783376	0.95

duckdb / vortex-file-compressed (0.927x ➖, 1↑ 0↓)

name	PR 277b386 (ns)	base 928738e (ns)	ratio (PR/base)
vortex_q00/duckdb:vortex-file-compressed 🚀	9998247	11407955	0.88
vortex_q01/duckdb:vortex-file-compressed	6070606	6193463	0.98

duckdb / parquet (0.963x ➖, 0↑ 0↓)

name	PR 277b386 (ns)	base 928738e (ns)	ratio (PR/base)
vortex_q00/duckdb:parquet	23461287	24228706	0.97
vortex_q01/duckdb:parquet	9394105	9799860	0.96

---

No file size changes detected.

[connortsui20]

see #8826 for related discussion on these semantics