bugfix in thirdDerivativeOfDistanceSquaredWRTFirst1AndSecond2Argument. Now it produces the same result as the fd approximation at least in for S^1 [[Imported from SVN: r7066]]
