Simulate memory queues in parallel for BWG

[0xVolosnikov]

What ❔

The greatest amount of time during BWG is spent simulating queues. At the moment, the queue simulation is single-threaded process. But different queues can be simulated in parallel, since they do not depend on each other.

This pull request does the following:

• Move simulation of memory queue implicit part out of individual_circuits. Currently when processing memory-related individual circuits, a part of the queue is simulated for each of circuits. But a more efficient approach is to first collect all the implicit queries and simulate the entire memory queue before memory-related circuits.
• Since we can now simulate the entire memory queue without interruption, including the implicit part, we also simulate a sorted version of the queue in parallel.
• Part of witnesses produced during simulation is sent to storage immediately (to greatly reduce RAM usage).
• Callstack simulation moved to separate thread.

As a result, a decrease in the time on the BWG by ~40% is observed locally.
Peak RAM decreased by ~30%

Checklist

• PR title corresponds to the body of PR (we generate changelog entries from PRs).
• Tests for the changes have been added / updated.
• Documentation comments have been added / updated.
• Code has been formatted via zk fmt and zk lint.

[0xVolosnikov]

Should be used together with matter-labs/zksync-era#2652