Clarify how to integrate ranger into the OrbitAI app

[georgeslabreche]

• This story is part of epic Evaluate the ranger library for batch training with Random Forest.
• This story is a prerequisite to Integrate ranger into the OrbitAI app.

Discussion Points

1. We can keep it super simple and just make it a binary that is triggered on demand via an NMF app action/command.
2. The binary would process the training data csv file.
3. We need to define what the interface is between the NMF app and the binary to allow for n-dimension training.
4. We would run the binary after manually checking that the training data log file has diverse enough data (i.e. for both 0 and 1 labels).
5. We should have a flag in the .properties file to indicate whether the app should log training data into a csv file. No need to persist data if we don't intended to use it.
6. We need to think of how we implement inference mode.

[georgeslabreche]

@TanguySoto:

1. For training: Define configuration properties for random forest training with ranger #51
2. For prediction: Implement on the fly prediction with ranger given a single data sample #52

Let's discuss so we can close this issue?

[georgeslabreche]

@TanguySoto refer to the new properties file:
https://github.com/georgeslabreche/opssat-orbitai/blob/ae4f34f0073618fb784c14e9e8cc2d1e72d9e630/nmf/space-app/conf/orbit_ai_v2.properties