feat: expose Spark scan schema and pushed predicates

[xiaoh1024]

Summary

Expose the scan metadata needed by alternative execution engines, such as Comet, to preserve filter semantics when replacing the Vortex Spark reader.

Spark may remove filters fully handled by Vortex and prune predicate-only columns from the scan output. The replacement reader therefore needs both the full table schema and the predicates the scan has accepted.

Changes

• Add VortexScan.tableSchema() to expose the schema before projection pushdown.
• Add VortexScan.pushedPredicates() to return a defensive copy of the accepted predicates. Document that replacement readers must enforce every predicate or retain the original reader.
• Add tests covering metadata preservation through filter pushdown, empty projection and scan construction, as well as defensive array copying.

Testing

• VortexScanTest: 2 tests passed on Spark 3.5.9 / Scala 2.12.
• VortexScanTest: 2 tests passed on Spark 4.1.2 / Scala 2.13.
• Java formatting checks passed.

AI Assistance

AI tools assisted with implementation, tests* and drafting this PR description.

[robert3005]

Just fyi I am working on #9658 which should simplify all of it

[codspeed]

Merging this PR will degrade performance by 9.3%

⚠️ 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.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 3 improved benchmarks
❌ 5 regressed benchmarks
✅ 1097 untouched benchmarks
⏩ 1366 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	decode_primitives[f32, (1000, 512)]	42 µs	64.8 µs	-35.09%
❌	WallTime	words_gather_dispatch_avx512[1024]	9 ns	13 ns	-30.77%
❌	Simulation	random_i16[0.95]	81.5 µs	99.8 µs	-18.36%
❌	WallTime	lanezip_checked_add_u32_avx2[16384]	5.3 µs	6.2 µs	-13.96%
❌	WallTime	dbp_split_kernel_mixed_null_neon[(I256, 1024)]	1.9 µs	2.2 µs	-11.65%
⚡	Simulation	random_i8[0.5]	96 µs	72.4 µs	+32.5%
⚡	WallTime	filtered_sink_i64_avx2[OneNullInEight]	26.1 µs	23.2 µs	+12.6%
⚡	WallTime	mul_u32_nonnull_avx512	6.2 µs	5.6 µs	+10.11%

Tip

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

---

_{Comparing xiaoh1024:feat/spark-native-scan-descriptor (6ab7fcf) with develop (895935e)}

Footnotes

1. 1366 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. ↩

[github-actions]

This PR has been marked as stale because it has been open for 14 days with no activity. Please comment or remove the stale label if you wish to keep it active, otherwise it will be closed in 7 days