Numpy wrap performance
We should look into numpy wraping, currently we are copying the data, which is not very nice.
Either we are able to use the jpype pointer and reinterpred data as numpy array or we make certain classes like the grids iteratable so we are able to apply python methods based on iterators.