Add Reasonable::Reasons for AI reasoning/thinking trace support

[TonsOfFun]

Summary

This PR adds support for capturing and tracking extended thinking traces from LLMs that support reasoning capabilities (like Claude's extended thinking and OpenAI o1 models).

New Components:

• SolidAgent::Reasonable - Concern for models to store reasoning data
• SolidAgent::Reasonable::Reason - Value object for reasoning traces
• SolidAgent::HasReasons - Concern for agents to capture reasoning
• Generator for adding reasoning columns to existing models

Features:

• Capture reasoning content and token counts from LLM responses
• Track thinking time and metadata
• Support for redacted reasoning (provider-redacted content)
• Reasoning chain aggregation for multi-step processes
• Statistics and summarization helpers
• Database persistence support via HasContext integration

Motivation

Inspired by the AgentFragment pattern in tonsoffun/writebook which tracks content transformations. This extends that concept to track AI reasoning/thinking traces for:

• Transparency: Understanding why an AI made certain decisions
• Debugging: Identifying issues in AI logic
• Auditing: Maintaining records of AI decision-making processes
• Cost tracking: Reasoning tokens are often billed separately

Usage

class AnalysisAgent < ApplicationAgent
  include SolidAgent::HasReasons
  include SolidAgent::HasContext

  has_reasons auto_capture: true, persist: true

  def analyze
    result = prompt(
      messages: analysis_messages,
      extended_thinking: true
    )

    # Access captured reasoning
    last_reasoning.content      #=> "Let me analyze this systematically..."
    total_reasoning_tokens      #=> 450
    reasoning_chain             #=> Full reasoning trace
  end
end

Test plan

• Unit tests for Reason value object
• Unit tests for HasReasons concern
• Full test suite passes (189 tests, 441 assertions)

🤖 Generated with Claude Code