Since the estimators return a StanFit object as a result of calling .fit() will they be compatible with sklearn pipelines?