v0.0.18

Release Notes - v0.0.18

Added deepseek provider and fixed initial_response type in callback

Added

• Deepseek LLM provider with two models:
  • deepseek-chat: Standard chat completion model
  • deepseek-reasoner: Enhanced model providing reasoning before answers
• Support for reasoning callbacks in Task system
• Improved response type handling in Task system

Improvements

• Enhanced callback system to support reasoning events
• Added response_type parameter to Task._direct_llm_call
• Removed redundant type manipulation from callback wrappers