Use a non-random "randomness" value for testing

[ed255]

Currently we are using a different random value on each test run for the randomness used for the RLC in the circuits:

zkevm-circuits/zkevm-circuits/src/evm_circuit/witness.rs

Line 1361 in 46dd697

randomness: Fr::rand(),

I believe it's better for tests to use always the same "random" value to get reproducible executions and run the test always with the same intermediate values. Even more, we can use a particular value that makes debugging an RLC encode value easier, like 0x1000 as suggested by @lispc .

See #622 (comment) for a discussion on this topic.

[CPerezz]

I would suggest to use SeedableRng

[CPerezz]

Well I was working on this but makes sense to use a fixed randomness value as basically it's for debugging purposes. And also, we don't have any Rng traits associated anywhere. So we're much more free in that regard.

Aside from that, even using in 0x1000 can be a solution now, we should keep an eye on this so that we have a solution that can scale with the codebase.

[CPerezz]

I think we can close this and open a new issue regarding the implementation of a global randomness solution for our testing.

[CPerezz]

Closing this in favour of #684