Non-hiding variant of Hyrax

[mmagician]

Summary

The current implementation of Hyrax in arkworks is closely following the PCS from Hyrax paper, which only presents the hiding variant. This means that the verifier never learns the value f(x) for the queried point x - he can only get convinced that the committer is able to open it (hence the hiding).

This is actually an undesirable property in some scenarios, as it leads to unnecessary work both for the prover and the verifier. This GH issue is for discussing whether:
a) the current implementation should be adapted to have optional hiding
b) create a separate scheme without the hiding part

Proposal

Since this should also serve as a learning exercise, I propose to start from a non-hiding variant from scratch - the scheme should actually be quite a lot simpler in this scenario.

Afterwards we should adapt our Hyrax benchmarks results