ST_Intersects pushdown: vortex.geo.intersects kernel, DuckDB lowering, spatial function overrides

[HarukiMoriarty]

Rationale for this change

ST_Intersects over native geometry columns currently evaluates inside DuckDB, exporting every row as GEOMETRY just to be tested. This adds a native vortex.geo.intersects kernel and pushes the filter into the scan: on SpatialBench SF1 (6M points vs a literal polygon), 406ms (DuckDB filter) / 36.6ms (SPATIAL_JOIN) -> 12.4ms pushed. Part of the native geospatial lane, following the ST_Distance/ST_DWithin pushdown.

What changes are included in this PR?

• vortex-geo: GeoIntersects kernel.
• vortex-duckdb: lower st_intersects(native column, GEOMETRY literal) like st_distance.
• vortex-duckdb: shadow spatial's ST_Intersects with a non-fallible copy — DuckDB won't push can-throw filters through the projection every view-registered table has. The ST_DWithin override generalizes into a table-driven registry (spatial_overrides.hpp) shared by registration and the join-condition restore pass.

Note: drop GeoDistance's columnar point fast paths. geo is row-oriented, so kernels materialize rows regardless; hand-rolled columnar paths belong in a future columnar geometry compute library (pushed Q1: 5.4ms -> 12.6ms, still far ahead of unpushed).

[codspeed-hq]

Merging this PR will not alter performance

⚡ 4 improved benchmarks
❌ 1 regressed benchmark
✅ 1625 untouched benchmarks
🆕 7 new benchmarks
⏩ 42 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	slice_empty_vortex	339.4 ns	397.8 ns	-14.66%
⚡	Simulation	chunked_varbinview_into_canonical[(100, 100)]	307 µs	271.9 µs	+12.91%
⚡	Simulation	encode_varbin[(1000, 8)]	160.7 µs	144.7 µs	+11.05%
⚡	Simulation	encode_varbin[(1000, 4)]	159 µs	144.2 µs	+10.32%
⚡	Simulation	encode_varbin[(1000, 32)]	164.7 µs	149.5 µs	+10.13%
🆕	Simulation	ilike_contains	N/A	783.5 µs	N/A
🆕	Simulation	like_contains	N/A	155.3 µs	N/A
🆕	Simulation	like_exact	N/A	43 µs	N/A
🆕	Simulation	like_per_row_patterns	N/A	167.3 µs	N/A
🆕	Simulation	like_prefix	N/A	49.8 µs	N/A
🆕	Simulation	like_regex	N/A	622 µs	N/A
🆕	Simulation	like_suffix	N/A	65.8 µs	N/A

Tip

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

---

_{Comparing nemo/geo-intersects (840b7ac) with develop (2a58c67)²}

Footnotes

1. 42 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. ↩

2. No successful run was found on develop (d2b2378) during the generation of this report, so 2a58c67 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[myrrc]

Let's add a sqllogictest which checks our queries don't break if we override the functions

[myrrc]

LGTM!