Validators circuit: trusted shuffling

Investigate the impact of not checking shuffling and committee assignments in `ValidatorsCircuit`.

The analysis must include:
- Attacks a malicious prover would be able to perform.
- Power (number of controlled validators, stake, compute power) needed to make a halt protocol execution and (if possible) forge the proof (construct an unsound proof).

> The problem space overview can be found in spec [here](https://hackmd.io/@timofey/ByTGrG9S2#On-shuffling-constraints).

Update documentation in spec and conclude whether avoiding shuffling constraints possible (at least as a temporary measure).

> **Note**: As an example, consider how in Scroll zkEVM precompiles are "trusted" (computed off-circuit but later used in proof generation using via lookup arguments).