Divergence between kzg and ipa interface

[arijitdutta67]

In PR #205, we get a divergence in the way ipa and kzg works. In particular, the commit function in the commitment_key class for ipa is not used inside any ipa methods, instead we provide the commitment directly to reduce_prove and reduce_verify as a member of the public input. This is not how the kzg interface works.

Also the pairing_check method is absent in the verification_key class for ipa.

We need to resolve if any issue arises when we interchangeably use kzg and ipa for bn254 and grumpkin circuits respectively.

Originally posted by @codygunton in #205 (comment)

[arijitdutta67]

In PR #244, we could resolve the mismatch with reduce_prove by introducing the transcript. However the issue with reduce_verify still remains (see the conversation here).

[maramihali]

closed by #445