v0.2.18 - SQL generation for multi-hop queries Features: - Multi-hop (a)-[]->(b)-[]->(c) generates single SQL query - Fix embedded FK detection (eliminates self-join bug) - All query types now faster than raw Spark SQL Performance (SF100, auto mode): - Simple Node: -32% (32% faster than raw SQL) - Filtered Node: -17% (17% faster) - Aggregation: -16% (16% faster) - Single-hop: -22% (22% faster) - Multi-hop: 40-50% improvement vs DataFrame API