feat: including the possibility to choose the hyperparameter of a given model.

[MarianoChaves]

Here I included the possibility to include the desiderd hyperparameters of a given class. It can be useful depending on the usecase. I've included it in the documentation and also included an unity test for that.

Disclaimers:

• Setting the hyperparameter is optional;
• If the hyperarameter is setted, it is not necessary to set to all models, only for the desired ones.

[shankarpandala]

Thanks for your contribution.
Will look into it

[shankarpandala]

Closing as this contradicts LazyPredict's core philosophy.

LazyPredict is designed for quick model comparison without hyperparameter tuning. The entire point is to provide baseline performance with default parameters.

If you need hyperparameter tuning, use dedicated tools:

• scikit-learn's GridSearchCV or RandomizedSearchCV
• Optuna
• Ray Tune
• Hyperopt

The workflow should be:

1. Use LazyPredict to identify promising models
2. Then tune the best models with dedicated tuning tools

Adding hyperparameter tuning to LazyPredict would defeat its purpose and make it slow.