Feedback on least-squares

[vcrispoli]

On the best-fit ellipse example there's an error on the last given point (-1,-1). In order to give the result presented the point should be (-1,1), as shown in the plots. Please correct :)

In reference to version 224734d