The Optex examples done here are noiseless. In fact, this is a misapplication of Optex. It is built on locally weighted learning methods that are made to handle noisy data, and it is most useful with noisy data.
The Optex examples here are one dimensional. These examples were chosen specifically because it is easiest to understand the methods on one dimensional problems. In particular, it is difficult to see confidence intervals for higher dimensional models. Optex works well in higher dimensional problems. It is probably more useful in those problems since they're harder to visualize and optimize by hand. In general, Optex will take longer to make its choices as the dimensionality and the number of data points goes up.