Improve iterative storage

Right now, the iterative storage scheme stores the results timestep by timestep. However, due to the nature of dask arrays, every timestep requires the computation of the block that contains the timestep. Therefore, storage can be improved by storing the results block-by-block instead.