memory efficient implementation with many instances

[dalessioluca]

right now big_mask and big_imgs have the spatial extension of the original image and an extra dimension for the number of instances. This will kill the memory for large images containing many cells. Is there a way to get away with this? It is interesting to note that the in the only the combination sum_j p_j m_j then I can drastically reduce the memory usage. What about imgs? Is the same thing true, i.e. sum_j p_j img_j?