Ensuring equal proportion of models, if multiple algorithms are provided

[haghish]

When calculating weighted mean SHAP values for multiple models, ensure that the models equally represent different algorithms, if models trained on more than one algorithm are presented. For example, if models from GBM and EXGboost are provided, ensure that the user is notified that the weighted mean SHAP values do not equally represent both algorithms, if models trained using one algorithm outnumber the other. Under such situations, either the user should be warned and optionally, the package should be allowed to select equal number of models from each algorithm to be more fair in its overall assessment of feature importance.