fix(ElemPerf): issue with contraction

The contraction should be calculated based on the **last element**
position and not the actual rotor radius.

In some edge cases, an element of the aft rotor can fall in between the
contraction calculated from the previous rotor's last element and the one
calculated from its tip radius. When this happens, the velocity is
poorly interpolated and brought to 0 for this element on the aft rotor.
This not only skews the results, but also lead to issues with the
current implementation of the indfact solver, which only looks into
Op.speed~=0 and into the airspeed for each individual elements.