Add support for interactive Agents

[fabecode]

Problem

Currently, the Agentic AI Framework supports autonomous agents that process documents and perform tasks independently. However, there's no built-in support for interactive agents that can engage in back-and-forth conversations with users or require human input during processing workflows.

Proposed Solution

Extend the existing Agent Framework to support interactive capabilities:

• Conversational interfaces - Enable agents to maintain context across multiple interactions
• Human-in-the-loop workflows - Allow agents to pause and request human input/approval
• Real-time communication - Support for chat-like interfaces (WebSocket, Server-Sent Events)
• Session management - Maintain conversation state and context

Use Cases

• Customer support chatbots integrated with document processing
• Interactive document review where agents ask clarifying questions
• Approval workflows where agents request human confirmation before actions
• Guided onboarding where agents walk users through complex processes

Acceptance Criteria

• New InteractiveAgent construct that extends base Agent functionality
• Real-time communication interface (WebSocket, SSE etc.)
• Configurable session management with pluggable persistence options
• Integration hooks for existing document processing workflows
• Example implementation demonstrating interactive capabilities
• Documentation covering architecture patterns and usage examples
• Testing including unit testing and NAG testing

Technical Considerations

• Should integrate with existing Agent Framework patterns
• Consider scalability for multiple concurrent conversations
• Security considerations for user sessions and data
• Cost optimization for long-running interactive sessions