Distributed iterative thresholding for ℓ0/ℓ1-regularized linear inverse problems