[ARCH] Replace flat registry with true graph data model (nodes + edges)

[Wolfvin]

Problem

Currently CodeLens uses a flat registry (SQLite rows per symbol). This works for simple lookup but blocks structural queries like:

• Who calls this function, across the entire codebase?
• What is the blast radius if I rename this class?
• Are there circular dependency chains?

Many existing engines (trace, impact, circular) reimplement partial graph traversal ad-hoc on top of the flat registry, leading to duplicated logic and inconsistent results.

Proposed Change

Adopt a proper node + edge graph model backed by SQLite:

Nodes: Function, Class, File, Module, Route, Type, Interface
Edges: CALLS, IMPORTS, DEFINES, INHERITS, IMPLEMENTS, USES_TYPE

Each edge stores: source_id, target_id, edge_type, file, line, confidence.

This enables:

• Unified traversal for trace/impact/circular/dead-code — all become graph queries
• Cypher-like query engine on top (see separate issue)
• Cross-file and eventually cross-repo analysis

Migration Path

1. Add nodes + edges tables alongside existing registry (non-breaking)
2. Populate during scan pass
3. Migrate engines one by one to use graph queries
4. Deprecate flat registry once all engines migrated

Inspiration

codebase-memory-mcp uses this model and achieves <1ms query time on 4.8M nodes via SQLite with proper indexing.

[Wolfvin]

Closing as duplicate of #8 (identical title, identical body, same intent — true graph data model with nodes + edges). Consolidating tracking there. If this was intentional, please reopen with what aspect is distinct.