v0.2.0 — GraphRAG, OpenCode & Smarter Caching

• GraphRAG: Builds a relationship graph between code entities to improve search retrieval
• OpenCode support: init now generates an opencode.json config
• The init CLI command now gives the option to start embedding immediately
• File locking: Prevents multiple CodeEmbed instances from processing the same file simultaneously
• Chunk caching: LLM results are cached per chunk — only changed chunks are reprocessed on re-embed, cutting redundant LLM calls
• Faster embed loop: Rechecks for file changes immediately after processing instead of sleeping