Add UnionArray compute functions

[connortsui20]

Rationale for this change

Depends on #8834. Tracking issue: #8769.

This is the deferred compute layer for UnionArray and remains a draft for later review.

What changes are included in this PR?

• Adds UnionArray filter, take, slice, and mask execution.
• Adds constant and chunked canonicalization.
• Adds Union casting, uncompressed-size accounting, and recursive compression.
• Adds focused tests for structural operations and canonicalization.

Nullable-index take remains explicitly unsupported and needs resolution before this layer is ready.

What APIs are changed? Are there any user-facing changes?

No additional public array representation APIs beyond #8834. This layer wires UnionArray into existing compute paths.

[github-actions]

Polar Signals Profiling Results

Latest Run

Status	Commit	Job	Attempt	Link
🟢 Done	ab33ec0		1	Explore Profiling Data

Previous Runs (6)

Status	Commit	Job	Attempt	Link
🟢 Done	f2427ce		1	Explore Profiling Data
🟢 Done	e32db9e		1	Explore Profiling Data
🟢 Done	98f8f06		1	Explore Profiling Data
🟢 Done	1e867dd		1	Explore Profiling Data
🟢 Done	a6b7e6c		1	Explore Profiling Data
🟢 Done	6fe53ba		1	Explore Profiling Data

---

Powered by Polar Signals Cloud

[github-actions]

Benchmarks: Vortex queries 📖

Verdict: No clear signal (low confidence)
Attributed Vortex impact: +4.5%
Engines: DataFusion No clear signal (+8.1%, low confidence) · DuckDB No clear signal (+1.0%, low confidence)
Vortex (geomean): 1.006x ➖
Parquet (geomean): 0.984x ➖
Shifts: Parquet (control) -1.6% · Median polish +1.2%

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 (1.040x ➖, 0↑ 0↓)

name	PR ab33ec0 (ns)	base 69f5bc5 (ns)	ratio (PR/base)
vortex_q00/datafusion:vortex-file-compressed	10068833	9640865	1.04
vortex_q01/datafusion:vortex-file-compressed	6306581	6087569	1.04

datafusion / parquet (0.962x ➖, 0↑ 0↓)

name	PR ab33ec0 (ns)	base 69f5bc5 (ns)	ratio (PR/base)
vortex_q00/datafusion:parquet	19895977	20804028	0.96
vortex_q01/datafusion:parquet	4487039	4638739	0.97

duckdb / vortex-file-compressed (1.017x ➖, 0↑ 0↓)

name	PR ab33ec0 (ns)	base 69f5bc5 (ns)	ratio (PR/base)
vortex_q00/duckdb:vortex-file-compressed	10131356	9980322	1.02
vortex_q01/duckdb:vortex-file-compressed	6014631	5899636	1.02

duckdb / parquet (1.007x ➖, 0↑ 0↓)

name	PR ab33ec0 (ns)	base 69f5bc5 (ns)	ratio (PR/base)
vortex_q00/duckdb:parquet	23421945	23225257	1.01
vortex_q01/duckdb:parquet	9419267	9364034	1.01

---

No file size changes detected.

[codspeed-hq]

Merging this PR will degrade performance by 14.37%

⚡ 1 improved benchmark
❌ 2 regressed benchmarks
✅ 1844 untouched benchmarks
⏩ 46 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	copy_nullable[65536]	579.1 µs	1,033.3 µs	-43.96%
❌	Simulation	copy_non_nullable[65536]	799.4 µs	913.8 µs	-12.52%
⚡	Simulation	decode_varbin[(1000, 8)]	34.3 µs	26.8 µs	+28.06%

Tip

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

---

_{Comparing ct/union-array-compute (ab33ec0) with ct/union-array (772b738)}

Footnotes

1. 46 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. ↩

[codecov]

Codecov Report

❌ Patch coverage is 74.09910% with 115 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.13%. Comparing base (cd92358) to head (98f8f06).

Files with missing lines	Patch %	Lines
...rtex-array/src/arrays/constant/vtable/canonical.rs	49.42%	44 Missing ⚠️
vortex-array/src/scalar_fn/fns/cast/mod.rs	0.00%	17 Missing ⚠️
vortex-array/src/arrays/union/tests.rs	93.59%	13 Missing ⚠️
...gregate_fn/fns/uncompressed_size_in_bytes/union.rs	0.00%	11 Missing ⚠️
vortex-array/src/arrays/masked/execute.rs	0.00%	10 Missing ⚠️
vortex-compressor/src/compressor/cascade.rs	0.00%	8 Missing ⚠️
vortex-array/src/arrays/filter/execute/mod.rs	0.00%	6 Missing ⚠️
vortex-array/src/arrays/dict/execute.rs	0.00%	3 Missing ⚠️
...aggregate_fn/fns/uncompressed_size_in_bytes/mod.rs	0.00%	1 Missing ⚠️
...ortex-array/src/arrays/chunked/vtable/canonical.rs	97.14%	1 Missing ⚠️
... and 1 more

☔ 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.