Created by: grygielski

This PR resolves problem with tests failing when compared values are large in magnitude because checking for absolute error between them is often wrong due to floating point calculation precision.