Is the instrumented-pointer method of calibrating anatomical landmarks in 3D motion analysis reliable?