derive PyTorchIETaskmoduleModelHubMixin from HyperparametersMixin

[ArneBinder]

This will allow to just call self.save_hyperparameters() in the __init__() of any Taskmodule (instead of passing all relevant parameters to super().__init__()) and config could be implemented (in analogy to PyTorchIEModel) as:

    def _config(self) -> Dict[str, Any]:
        config = dict(self.hparams)
         config["taskmodule_type"] = self.__class__.__name__
        return config