feat: extend repository intelligence with symbol calls and test impact v2

[Mopati123]

Purpose

Extend the certified repository-intelligence v1 baseline into v2 with symbol-level call relationships, test-to-code mapping, and component-aware change impact.

v2 additions

• conservative Python symbol-call graph
• test-to-code dependency edges
• top-level repository component mapping
• impacted test identification
• impacted component identification
• impact reasons distinguishing changed/import/test relationships

Architectural role

This moves the repository-intelligence layer closer to the governed Developer OS requirement:

change request
  -> changed files
  -> impacted files
  -> impacted symbols
  -> impacted tests
  -> impacted components
  -> bounded OpenHands read/test/patch scope

Truth boundary

This remains deterministic static analysis. Dynamic imports, reflection, runtime dependency injection, generated code, monkey-patching, and broad cross-language semantic calls are not claimed as covered.

Review order

This PR is stacked on the certified v1 branch and should remain draft until its dedicated CI passes.

Certified v2 baseline

At head e90ef4db2ec6e0a8aba867124bbe4a4f81afd5a1, the dedicated Repository Intelligence CI completed successfully.

Certified v2 additions:

• conservative Python symbol-call graph;
• test-to-code dependency edges;
• top-level component mapping;
• impacted test identification;
• impacted component identification;
• deterministic impact reasons.

Truth boundary remains static-analysis only: dynamic imports, runtime reflection/injection, generated code, monkey-patching, and broad cross-language semantic calls are not claimed as covered.

This commit is the certified Phase M v2 baseline. PR remains draft and unmerged.