Use numpy arrays instead of Python lists for working with data in-memory

[elfprince13]

Besides the performance benefits (and interoperability benefits for client code), this will obviate the need nestedListToBin as a byte-buffer view of the array can be obtained with the .data attribute. Additionally, numpy arrays are endian-aware, so we can still check if the array is big endian and get a vectorized conversion with .newbyteorder if needed.

May be helpful if paired #13.