Implement Phase 2: markdown node model and graph index

[AlexMercedCoder]

Summary

Implements Phase 2 — Markdown node model & index from the implementation plan.

Node model (maggraph::node)

• YAML frontmatter parser → typed NodeMetadata (id, type, source, links, plus extra fields via #[serde(flatten)])
• Node struct with body content and relative path
• Markdown round-trip serialization that preserves unknown frontmatter keys
• Node::from_file / Node::write_to for filesystem I/O

Graph index (maggraph::index)

• GraphIndex::open(root_path) — full scan of *.md files under the graph root
• In-memory maps: id → NodeIndexEntry and path → id
• Duplicate ID detection on scan
• CRUD: create_node, read_node, update_node, delete_node, rescan

Other

• Extended MagGraphError with node/index variants
• Added serde_yaml and walkdir workspace dependencies
• PRD example node at examples/lakehouse/knowledge_graph/customer_churn_q2.md
• Updated planning/PROGRESS.md — Phase 2 marked complete

Testing

cargo test
cargo clippy --all-targets --all-features -- -D warnings

20 unit tests (12 new) covering PRD schema parsing, round-trip, scan, CRUD, duplicate detection, and the basic example fixture.

Next up

Phase 3 — wikilink parsing, adjacency, and traversal API.

Slack Thread