Currently, Df uses its own type Data, which is isomorphic to Maybe. This is very inconvenient, because we need convert between these types via Df.dataToMaybe and Df.maybeToData every time we need a function that specifically works on Maybe.

The original motivation for this design choice was that Maybe is lazy, but that does not seem to be a problem anymore. Moreover, other protocols such as Axi4Stream and PacketStream also use Maybe to indicate validity.

We can also use https://hackage.haskell.org/package/strict-0.5.1/docs/Data-Strict-Maybe.html, but that does require an extra dependency.