Improve five-language code graph quality

[forhappy]

Summary

• improve default-low Python import, receiver, singleton, alias, and provenance resolution
• preserve exact Rust reexport precedence, Go closure return semantics, Java source-set detection, and TypeScript callable-property targets
• improve source-backed natural-query ranking while retaining genuine ambiguity and bounded store-backed discovery
• harden the qualification timing parser and update the compatibility, implementation, and semantic-evidence contracts

Qualification result

On source-reviewed positive questions across Python, Rust, Go, Java, and TypeScript, Compass answered 100/100 top-1 correctly versus Graphify 14/100 under the same default 2,000-result cap. Both passed 25/25 negative controls. Build time and query latency were recorded but not used as quality criteria.

Validation

• cargo fmt --all -- --check
• cargo clippy --workspace --lib --bins --locked -- -D warnings
• cargo test --workspace --lib --bins --locked
• cargo test -p compass-files -p compass-languages -p compass-model -p compass-query -p compass-resolve --lib --tests --locked
• python3 -m unittest benchmarks.performance.tests.test_analyze
• sh scripts/check_product_boundary.sh
• ./scripts/qualify_code_graph_v1.sh --fixtures-only

All compiling Cargo commands used Rust 1.97.1 and an external per-worktree target directory under /Volumes/Workspace/crabbuild-target.