v0.2.0

• added support for markdown tables in documentation parsing
• improved answer accuracy with context window expansion to 8k tokens
• fixed crash when processing empty code blocks
• streaming responses now enabled by default
• refactored embedding pipeline to reduce memory usage by ~40%