Same as Fig. B.3, now for bias:800Ares2/recovery:800Ares2, i.e. the high-resolution setup, and considering systematic-tracers rather than PCA components. Nearly perfect one-point correlation corrections (left) are accompanied by the cross-correlation between unbiased FLASK data and organised randoms (right; orange) rising to ∼50% of the unbiased clustering signature, compared with the negligible cross-correlation in Fig. B.3.