Enforcing evaluations

[mimoo]

@mrmr1993 keeps finding missing evaluation proofs in PRs that add evaluations to the proof. It is not easy to change the protocol safely, and there are some obvious places that should always be checked. The question is, how can we modify the implementation and, perhaps leverage the type system, to ensure that no evaluations will ever be missed?

[mimoo]

same as with absorbing all of the evaluations before the evaluation proof (cf #211 #221)