Implement Programmatic FLEE Execution

[nicleosch]

Preliminary Information

• This pull request originates from collaborative efforts between students from the Technical University of Munich and the World Food Programme (WFP). The objective is to forecast the movements of Internally Displaced Persons (IDPs).

Summary

This pull request introduces a new module, runner.py, containing a class, Simulation, that allows flee execution via function calls extending the previously used command line interface. The goal is to provide developers with a programmatic way to interact with flee within their applications.

Changes Made

• Added a new module: runner.py
• Implemented functions for flee invocation, providing a more flexible and modular approach
• Refactored existing code to accommodate the new module and enhance maintainability
• Instead of writing a file, the output is a python dictionary containing the simulation results
• The simulation results includes refugees/IDPs at every location, providing valuable insights beyond camp-specific data

How to Use

Developers can now utilize the runner.py module in their codebase to programmatically invoke the tool. Example usage:

from runscripts.runner import Simulation

# Create an instance of the Simulation class
sim = Simulation(input_dir, validation_dir, duration, simsettings)

# Invoke flee using the function call
output = sim.run()

# Process the output as needed

Benefits

• Easier integration of flee into existing codebases
• Improved flexibility for developers to customize flee invocation

Testing

• Unit tests cover various scenarios, including edge cases and typical usage patterns.
• feel free to write additional tests

Reviewer Checklist

• Verify the correctness of the new module and its functions.
• Check for adherence to coding standards and best practices.
• Review test coverage and ensure appropriate testing scenarios.

[djgroen]

Okay. Overall I think this is a sensible change, although the contents of runner.py does overlap strongly with those of run.py and run_par.py, which are already widely used.

This is not a serious issue, but it just means that whenever we update the run functions, we will have to update this now in three places, not two, to prevent things from breaking.

Also, I think it would be good to mention this new functionality somewhere in the docs/ documentation. Perhaps it's worth adding that?

Lastly, I see that some of the tests are failing, but if they work then I'm happy to merge this in.