More descriptive and easier abstraction for recursive proofs generation

[signorecello]

Part of the first noir_js devex review.

To prove, we call prover.generateOuterProof which was a temporary name @kevaundray came up with for “generate a proof for some circuit not meant to be verified in another circuit”.

Problem is, it's not so descriptive and we could try to abstract that away or at least make it easier to understand.

Some suggestions:

Flag on the constructor

Noir.js could get the "type of proof" from the constructor, and expose just one function prover.generateProof, example:

const noir = new Noir(circuit, { recursive: true })
const proof = noir.generateProof() // noir.js already knows if it's dealing with a recursive circuit

Define new API

We could define a new API just for recursive proofs, possibly inheriting some base class:

import { Noir, NoirRecursive } from "@noir-lang/noir_js";

const noir = new Noir() // non-recursive instance of Noir
const recursive = new NoirRecursive() // recursive instance of Noir

Rename methods

This is probably the easiest: just renaming the methods. @Savio-Sou came up with some cool alternative names to generateOuterProof:

const noir = new Noir();
const proof = noir.generateIntermediateProof() // only people who care about recursive proofs know what does this mean
const proof = noir.generateFinalProof() // descriptive enough for those who have intermediate proofs, but doesn't prompt any confusion to the common app developer