Pedersen cleanup

[kevaundray]

Problem

Now that Pedersen has been merged, the codebase is a lot cleaner however we still have a bit of tech debt related to the old pedersen implementations.

This is a tracking issue to note down all of these tech debt tasks

[kevaundray]

The first is use of compress and the plookup prefixed function names

#3033 removes plookup from the namespace
#3038 removes compress from the interface

[kevaundray]

We also have c_bind and c_bind_new -- since the generator runs on c_bind_new, we can get rid of c_bind and parties which rely on c_bind should switch to that calling convention

[kevaundray]

General investigation of hash/commit methods which are not used

[kevaundray]

We have a pattern of exposing hash methods which can be implemented by other exposed hash methods. One example is plookupCompress and PlookupCompressWithHashIndex, where the former is the latter with an index of 0

#3038 wittles down the interface to only have HashWithHashIndex and Hash

[kevaundray]

This can be a separate issue.

Investigate implementation of hash_buffer:

template <typename Curve>
typename Curve::BaseField pedersen_hash_base<Curve>::hash_buffer(const std::vector<uint8_t>& input,
                                                                 const GeneratorContext context)
{
    std::vector<Fq> converted = convert_buffer(input);

    if (converted.size() < 2) {
        return hash(converted, context);
    }
    auto result = hash({ converted[0], converted[1] }, context);
    for (size_t i = 2; i < converted.size(); ++i) {
        result = hash({ result, converted[i] }, context);
    }
    return result;
}

Intuitively, I would have expected it to convert a buffer into fields, then call hash with those fields and the context.

EDIT: this was is actually better because we re-use the same generators, instead of having N different generators

[kevaundray]

check for stale comments/TODOs in some of the pedersen files, the most obvious would be to grep for delete this file

[kevaundray]

Investigate whether we should get rid of pedersen__init -- it currently does nothing in the cbinds

[kevaundray]

Marking this as completed, #3076 and #3077 have been spun out into their own issues for further discussion