batch open multiple opcode polynomials.

[kunxian-xia]

Currently for each opcode circuit, we commit to several (less than 30) witness polynomials which has same number of variables. These polynomials are committed in one merkle tree and then open at a random point. Let's say If we include 10 opcode proofs inside a shard proof, then we need to commit to 10 merkle trees and generate 10 opening proofs. This incurs really big overhead.

As a comparison, STARK protocol would build only one polynomial for running low-degree testing subprotocol like FRI.