feat: Add Reactive LLM Opinion Dynamics example (Mesa 4.0 Clean Slate)

[ApurvaPatil2401]

Summary

Introduces a new example model: Reactive LLM Opinion Dynamics. This model explores how agents can utilize LLMs for reasoning in a performance-optimized, reactive manner.

Motive

The goal of this feature is to demonstrate a "Reactive Trigger" architecture. Traditional LLM-based agents often suffer from high latency and API costs; this model shows how to invoke reasoning only when specific environmental thresholds are met, significantly improving model efficiency while maintaining high cognitive fidelity.

Implementation

Built using Mesa 4.0 "Clean Slate" standards:

Agent-Centric logic: Moves away from the legacy centralized scheduler in favor of explicit triggers.

AgentSet.do(): Utilizes the new batch execution pattern for handling reactive opinion updates.

Forward Compatibility: Structured to eventually integrate with ContinuousSpaceAgent once experimental spatial APIs are stabilized.

Usage Examples

The model can be run headlessly to observe the reactive triggers in the terminal:

  python examples/reactive_llm_opinion/model.py

(A Solara visualization component is also included in app.py for future expansion.)