Today I remade the experiment of autorotation with only one other wing then I obtained the same result as on the following video I put again:
4-5 kg regular traction measured with a steelyard on the line taut with 4 m/s wind speed while I obtained 3 kg surge traction in crosswind larger figure-eight and for one of the two lines, the other line giving a lower value in the same time, and the total value being irregular and probably lesser.
So it looks like an autorotation with low radius loop could generate more power and using less space than crosswind large figure-eight that are generally envisaged.