Restructure ridge test for different dataset cases

[agoscinski]

We have certain tests in TestRidge which only work with when n_features << n_samples, so we have two data sets we are testing on. At the moment this is not so visible in the code, because we overwrite the same parameter names. We could choose different names for the parameters or pack them in a dictionary with different names. We could also use pytest fixtures for this which has the additional benefit to reduce code repetition since we do almost in every test:

        # Create input values
        X_arr = self.rng.random([2, num_targets, num_properties])
        y_arr = self.rng.random([2, num_targets, 1])
        alpha_arr = np.ones([2, 1, num_properties])
        sw_arr = np.ones([2, num_targets, 1])