More Learning fixes

[MahdiAll99]

This pull request introduces enhancements to how pipelines are handled and displayed, as well as improvements to model metadata management and presentation. The main changes include adding a method to split pipelines with complex keys, refactoring model handler initialization, ensuring the model algorithm is consistently saved and displayed, and updating the UI to show the algorithm used.

Pipeline handling improvements:

• Added a new split_pipelines method to MEDexperiment.py that processes pipeline keys containing dots (.), splitting them into separate entries for better handling of complex pipeline structures. This method is now called during pipeline initialization. [1] [2]

Model handler and metadata management:

• Refactored ModelHandler.py to move the logic for extracting and storing model metadata (such as model_id, model_name_id, and estimator settings) so it only executes for nodes of type train_model, preventing redundant or incorrect assignments for other node types. [1] [2]
• In ModelIO.py, ensured the model's algorithm name is consistently determined and saved in the model's metadata, and added it to the output dictionary under the algorithm key. [1] [2]

UI enhancements:

• Updated the model viewer UI (modelViewer.jsx) to display the machine learning algorithm used by the model, in addition to the type, if available.

These changes improve the flexibility of pipeline configuration, ensure accurate model metadata, and enhance the clarity of information presented to users.