feature - implement RFC 020 nested scalar functions

[dannymeijer]

Summary

Implements RFC 020 nested data functions as scalar, registry-backed InQL helpers for arrays, maps, and named structs. The branch keeps relation-cardinality-changing generator behavior reserved for RFC 021 while adding concrete Substrait mappings and DataFusion-backed execution coverage for nested scalar projections.

Type of change

• Bug fix
• New feature
• Refactor / maintenance
• Documentation
• CI / tooling
• RFC (adds/updates docs/rfcs/*)

Area(s)

• Package & tests
• Specification (RFCs)
• Documentation
• Automation & repo config
• Other

Key details

• User-facing behavior: Authors can build and inspect nested scalar values with helpers such as array(...), cardinality(...), array_contains(...), arrays_overlap(...), element_at(...), array_flatten(...), map_from_arrays(...), map_extract(...), map_contains_key(...), map_keys(...), map_values(...), map_entries(...), and named_struct(...).
• Internals: Adds a src/functions/nested/ helper family using declaration-side registry metadata, stable Substrait function anchors, concrete extension mappings where supported, and a documented map_contains_key(...) rewrite over existing scalar applications.
• Source quality cleanup: Nested helpers call registered_application(...) directly instead of routing through a one-line wrapper.
• Registry guardrail: Updates make registry-metadata to follow @register_function(...) decorator metadata, derive canonical names from checked callable metadata by default, and honor explicit canonical_name overrides. This prevents the guardrail from passing with zero helpers.
• Risks: Nested function semantics depend on backend nested UDF support. The PR documents one-based indexing and current recoverable element_at(...) adapter behavior instead of pretending a strict/try split exists yet.

Merge Order

#44 and #45 are merged. This PR targets main directly and can be reviewed as the RFC 020 follow-up.

Testing / verification

Post-#45 retarget verification:

• make fmt INCAN=/Users/danny/Development/encero/incan/target/debug/incan
• incan test tests/test_nested_data_functions.incn (3 passed)
• incan test tests/test_function_registry.incn (13 passed)
• incan test tests/test_substrait_plan.incn (21 passed)
• incan test tests/test_session_projection.incn (10 passed)
• make fmt-check INCAN=/Users/danny/Development/encero/incan/target/debug/incan
• make test-style
• make registry-metadata INCAN=/Users/danny/Development/encero/incan/target/debug/incan (78 helpers)
• make build INCAN=/Users/danny/Development/encero/incan/target/debug/incan
• make test INCAN=/Users/danny/Development/encero/incan/target/debug/incan (127 passed)
• make smoke-consumer INCAN=/Users/danny/Development/encero/incan/target/debug/incan
• make pre-commit INCAN=/Users/danny/Development/encero/incan/target/debug/incan (127 passed)

After source-quality cleanup:

• incan test tests/test_nested_data_functions.incn (3 passed)
• incan test tests/test_function_registry.incn (13 passed)
• make pre-commit INCAN=/Users/danny/Development/encero/incan/target/debug/incan (127 passed)
• make smoke-consumer INCAN=/Users/danny/Development/encero/incan/target/debug/incan

Docs impact

• No docs changes needed
• Docs updated
• Docs follow Divio intent (tutorial/how-to/reference/explanation) where applicable

If docs updated:

• Link(s): docs/language/reference/functions/nested.md, docs/language/reference/functions/index.md, docs/release_notes/v0_1.md, docs/rfcs/020_nested_data_functions.md

Checklist

• I kept public docs user-focused and moved internals to contributing docs when appropriate
• I avoided duplicating canonical install/run instructions in multiple places
• I added/updated tests where it materially reduces regressions

Closes #37