amesa-inference-dev

Create a mini-application that simulates a simple agent-based model using the 'amesa-inference-dev' Python package. This application will showcase the package's ability to perform inference on agents using ONNX models without requiring Ray or PyTorch. The app should include the following features:

1. **Agent Initialization**: Users should be able to define a set of agents with initial conditions such as position, velocity, and state.
2. **Model Inference**: Utilize the 'amesa-inference-dev' package to apply pre-trained ONNX models to each agent, updating their states based on the model's predictions.
3. **Visualization**: Implement a basic visualization tool to display the movement and state changes of the agents over time. This could be done using matplotlib or a similar library.
4. **Parameter Tuning**: Allow users to adjust parameters like the learning rate, model input dimensions, and agent interaction rules to observe different behaviors.
5. **Scenario Creation**: Provide templates for different scenarios such as predator-prey dynamics, flocking behavior, or disease spread among agents.
6. **Logging and Reporting**: Include functionality to log the simulation data and generate reports summarizing the outcomes.

The goal is to create an engaging and educational tool that demonstrates the power of agent-based modeling and inference using ONNX models. Focus on making the code modular, well-documented, and easy to extend for future enhancements.