Stateful, incremental API for VID

[ggutoski]

As a downstream consumer of VID, https://github.com/EspressoSystems/espresso-sequencer requires an incremental API so that txs can be added to a block one-at-a-time.

Change the VidScheme trait to give it a stateful, hash-like API (update, finalize, etc).

Update phase

• State: a list of txs.
• Caller can call update to add txs but cannot query anything.
• update input: individual tx payload serialized to [u8].
• Convert to Vec<F> via bytes_to_field_elements from jellyfish.
• Append to tx list.

Finalize phase

• State: KZG commitments, a list of txs & their indices.
• Caller cannot call update but can now query everything.
• Computationally intensive: construct polynomials, compute KZG commitments (requires FFT).
• Use the tx list to support proofs for individual txs.
  • Example: indices[i] is the polynomial input for the ith tx. A proof for the ith tx is a batch KZG opening for input puts indices[i]..indices[i+1]
  • Out of scope for this issue. See EspressoSystems/jellyfish#360 .

[ggutoski]

no longer relevant