Properly generalize across curve groups

[AaronFeickert]

The library currently generalizes its operations over groups, but not completely. Specifically, it generalizes group elements, but always assumes a scalar field compatible with Ristretto. This is very limiting and not a consistent abstraction.

There are a few options on how to proceed:

• Remove the abstraction, and limit the library to use of the Ristretto group
• Update the abstraction to handle a group's scalar field
• Overhaul the abstraction to use the group and ff traits, which are supported by the Ristretto library implementation

I favor the third option, as using the existing traits seems like the cleanest and most idiomatic approach.