We should try to follow the principle of keeping IO separate and cordon off the IO aspects of this package into its own module.

There could be an abcdmicro.io module that contains all the IO stuff -- specifics about how we read and write data to disk, governing which file formats we can handle.

We can discuss more below about how to approach this.

• Come up with a design for how to separate IO #44
• Implement and unit test the separation of IO (this issue), so that existing pipeline steps no longer explicitly do IO in their code.