optional contiguous memory handling for NumPy buffer access
There is a straight forward way of handling grid data (std::vector
hmm. I dislike exposing these kind internal details to the outside world. problem is that ilwis-objects for obvious reasons ( memory limitations) doesnt use contiguous memory. So exposing it is only a partial solution that will fail in a number of cases.
a Python API solution would be a derived BlockIterator that knows about the internal memory block size, which is then used in combination with numpy's direct memory access