A note on wavelet density deconvolution for weakly dependent data