Re-add add_constant_witness method for the composer.

[CPerezz]

The method was removed in #278 and there were no issues/actions which support this.
See: a8d8f3e#r41368517

The method was useful and is used across a lot of repos in the v_0.2.6 so I think we should bring it back. Idk if it was an error or not, but definitely should still be in the codebase.

[kevaundray]

Hi,

This method was introduced in 564e969

I believe this method is/will cause a bug;

• The method states that you are constraining a public input to a witness. However, it is constraining a selector polynomial value to the witness.

• There is already an identical method named: add_witness_to_circuit_descriptionwhich does the same thing.

The consequence of using add_constant_witness and believing that you are constraining public inputs, will be that each statement will create a new circuit. This is because selector polynomials define the circuit.

Solution

If you need to constrain a public input, I suggest a new method called constrain_public_input or add_public_input or something more similar. Then exposing that instead.

[kevaundray]

Could you possibly link the places which were using this method, so that I can double check the usage later on?

[CPerezz]

The goal of the function is to reduce this verbosity:
https://github.com/dusk-network/Poseidon252/blob/88169b39ccd36d017e08d7f5d94a8f7c82d5ca21/src/merkle_proof/proof.rs#L24
Or the most explicit case: https://github.com/dusk-network/Hades252/blob/7ba70c7b84504b9bc2df3e5c3ce8dbb8f02a3535/src/strategies/gadget.rs#L102

And instead have something like:
https://github.com/dusk-network/plonk_gadgets/blob/b208fc49703e67170356ccd88a81a45ffbad9182/src/gadgets/scalar/scalar.rs#L38

[CPerezz]

@kevaundray was right and there's a method that does the exact same thing called: add_witness_to_circuit_description but it asks fora variable and a Scalar when it could just ask for a Scalar and return the Variable. So I'll create an issue in order to address that.