The PR #68 implements the `Plan.execute` function which takes the training data, model params and client config as parameters.
