chore - speed up generated-project test suite

[dannymeijer]

Summary

This PR cuts the local 0.3 test/pre-commit cost by consolidating duplicate generated-project tests, folding LSP coverage into the primary nextest run, and adding a safe shared Cargo target cache for generated incan run / binary incan build test helpers. The warm verified full make pre-commit wall time is now 133.33s, with nextest at 101.002s and the full pre-commit gate core at 103s.

Type of change

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

Area(s)

Select the primary areas touched (used for review routing; labels are managed separately):

• Incan Language (syntax/semantics)
• Compiler (frontend/backend/codegen)
• Tooling (CLI/formatter/test runner)
• Editor integration (LSP/VS Code extension)
• Runtime / Core crates (stdlib/core/derive)
• Documentation

Key details

• User-facing behavior: No language behavior changes. Contributors get a faster local gate, and LSP tests now run as part of the main test command instead of a separate focused pass.
• Internals: Consolidates repeated runtime/generated-project test scenarios; adds INCAN_TEST_SHARED_TARGET_DIR for generated test harnesses; adds INCAN_GENERATED_CARGO_TARGET_DIR plus hashed binary target names for shared generated Cargo target reuse; keeps generated library target names stable for native Rust consumers.
• Risks: The main risk is cache aliasing across generated projects; binary target names are output-dir-hashed under the shared target override, and focused native-consumer/artifact tests verify that generated library crate names remain stable.

Testing / verification

• make test / cargo test
• make examples (if relevant)
• incan fmt --check . (if relevant)
• Manual verification described below

Manual verification notes:

• cargo test --no-run --lib — passed
• cargo test --no-run --test integration_tests — passed
• cargo test --lib shared_target_safe_name -- --nocapture — passed
• cargo test --lib relative_target_dirs -- --nocapture — passed
• cargo test --test integration_tests codegen_tests::test_run_c_import_this -- --nocapture — passed
• cargo test --test std_encoding_algorithm_modules std_encoding_algorithm_vectors_and_invalid_cases -- --nocapture — passed
• cargo test --test cli_integration run_accepts_generic_rust_param_scenarios_share_one_generated_project -- --nocapture — passed
• cargo test --test generated_rust_native_consumer_tests native_rust_consumer_can_call_generated_public_items -- --nocapture — passed
• cargo test --test generated_rust_artifact_tests generated_application_artifact_matches_baseline -- --nocapture — passed
• cargo test --test generated_rust_callability_artifact_tests generated_callable_artifact_and_consumers_share_producer_build -- --nocapture — passed
• make fmt — passed
• git diff --check — passed
• make pre-commit — passed; nextest ran 2442 tests in 125.182s, pre-commit-full-gate total was 164s, and smoke-test-fast passed
• /usr/bin/time -p make pre-commit — passed outside the sandbox; nextest ran 2442 tests in 101.002s, pre-commit-full-gate total was 103s, smoke-test-fast passed, wall time was 133.33s

Docs impact

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

If docs updated:

• Link(s): n/a

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