feat: agent todo list

[mhordynski]

Feature description

Implement todo list functionality that allows agents to break down complex tasks into smaller subtasks and track their progress (pending, in-progress, done).

Motivation

Agents often lose track of subtasks when solving complex problems. A ToDo list would help them work systematically, maintain context, and ensure all tasks are completed.

Additional context

Example usage:
  agent.create_todo_list({
      1: "Analyze current implementation",
      2: "Identify improvements",
      3: "Refactor components",
      4: "Write tests"
  })
  agent.mark_task(id: int, status: literal["in-progress", "done"])

Evaluate by running baseline without the todo list for task accuracy (for example coding), and another one with todo list enabled - measure improvement / loss.