slow performance

[Russell-Ryan]

There are a host of lines that use np.floor(), np.amax() and so on. Each of these lines has an embedded for-loop within numpy, and if the size of the input polygon is large or there are many polygons (even small ones), this can lead to significant performance problems. This can be fixed by using a JIT compiler (like numba) to perform all of these operations "in one go", considerably improving performance on large datasets. This was discovered by Massimo Griggio