feat: Implement GDS (Graph Data Science) Extension

[longbinlai]

What problem does this solve?

Implement the GDS (Graph Data Science) extension for NeuG based on the design specification in PR #64 (feat: Add GDS Design Spec).

This issue tracks the implementation of graph algorithm capabilities in NeuG, enabling users to:

1. Install/Load extensions - INSTALL EXTENSION gds, LOAD EXTENSION gds
2. Project subgraphs - CALL project_graph(g, {node_spec}, {edge_spec}) - zero-copy graph projection
3. Run graph algorithms - CALL algo_name(g, params) YIELD columns

Scope

Extension Infrastructure

• Extension registry and lifecycle management
• Extension build system (shared libraries)

Graph Projection

• ProjectedSubgraph data structure (zero-copy, store labels + predicates)
• Session-level subgraph context
• CALL project_graph implementation

Graph Algorithms (8 total)

• PageRank, Connected Components, Label Propagation, Leiden
• Louvain, Weakly Connected Components, Shortest Path, BFS

Integration

• GDSAlgo physical plan operator
• Cypher parser support for CALL ... YIELD ... syntax
• Integration with existing execution pipeline

Design Spec

See specs/004-gds/spec.md (1120+ lines, from PR #64) for full design details.

Related

• Design Spec: PR feat: Add GDS Design Spec #64 (feat: Add GDS Design Spec)
• Spec Author: @shirly121 (Xiaoli Zhou)