Flaky behavior of the tests depending on visualization option

[CaelmBleidd]

Description

Several tests for wrappers fail (or pass) the pipeline depending on the visualisation option value in UtSettings file. If we have this option enabled, everything works fine. If we disable it -- several tests fail.

To Reproduce

Steps to reproduce the behaviour:

It is flaky behaviour and there is no common way to reproduce it. Probably, it depends on the state of the system.

Visual proofs (screenshots, logs, images)

There are no specific visual proofs.
For example, org.utbot.examples.collections.MapEntrySetTest#testIteratorNext fails with absent second branch that leads to the NoSuchElementException.

Environment

No specific environment is required.

Additional context

I've investigated the problem recently. It happens because of the non-deterministic behaviour of the method peekImpl in NonUniformRandomSearch method.

When we have enabled visualization, it calls peek several times before each poll call to visualize the graph, which changes the next chosen state (because of the random usage in the method). At the same time, if we disable visualization, there will be no peek calls, and, therefore, we might have a different sequence of the chosen states from the execution queue.