[SD-154] Create a PyTorch energy estimation CLI tool

[dudeperf3ct]

Few changes done to the model_training

• Issue with mlflow loading the saved model using mlflow.pyfunc.load_model (No module named model_builder). Since we are using a custom sklearn pipeline, we have to save the model definition as well (how to load the saved model).
• For resolving above mlflow issue, we manually log the model instead of relying on autologging and use code_paths to add model_builder folder while logging the model.
• Additionally, trainer.py is moved as mlflow expects a particular format for saving the model using code_paths argument. See here for more details.

[dagshub]

Join the discussion on DagsHub!

[OCarrollM]

looks good to me! nice work