[ML-aware Architecture] Definition of the ML pipeline

[fwilhelmi]

In order to execute the ML functionalities at the corresponding nodes/entities, the ML pipeline must be defined. Moreover, we must provide a clean separation of ML functionalities to be executed through the ML pipeline.

In Komondor, the ML pipeline will be defined at both agents and central controller, where different functionalities may be activated according to the use case.

[fwilhelmi]

To do this, we need to generate abstract classes (e.g., pre-processor) and link them with the corresponding elements (e.g., AP) through pointers.