The current public facing API:

Should we:

• add methods to the extended data square? For instance: serialization and deserialzation (e.g. from and to protobuf), (square/orig. data) Width
• clarify the names of the existing Encode/Decode functions (e.g. by exposing the methods of the Codec interface and removing the existing Encode/Decode functions)
• decouple the erasure coding step from the row/col roots step (that way the library can be used with greater flexibility)

related: #11