Move generic information to a state memory

[CaelmBleidd]

Description

The initial issue occurred since we used to have path-independent generics: if during the analysis we discover, that our class is a subclass of another one, and now we might have other generic info (including a number of type arguments), we would take into account generic information for a random path and apply it for each of them.

During implementation I discovered that there are more problems, e.g., if we have several consecutive generic info in one jimple instruction that should be applied for the same object, the latest one will override the earlier one even if it consists of less information.

Moreover, there was an error with retrieving generic type info with upper type bounds, because of which we created type storage with the wrong leastCommonType. It could cause an error during set or get operations for RangeModifiableArray since it uses only a leastCommonType to construct type storage for an object it works with. Because of that, in fact, we didn't use generic types in many situations.

Fixes #1668

Type of Change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

Regression and integration tests

They are the same as automatic tests. But it can be checked using the ContestEstimator: there are no errors for the example from issue #1668

Automated Testing

org.utbot.examples.types.PathDependentGenericsExampleTest#testPathDependentGenerics
org.utbot.examples.types.PathDependentGenericsExampleTest#testFunctionWithSeveralTypeConstraintsForTheSameObject

Manual Scenario

There are no specific manual scenarios.

Checklist:

• The change followed the style guidelines of the UTBot project
• Self-review of the code is passed
• The change contains enough commentaries, particularly in hard-to-understand areas
• New documentation is provided or existed one is altered
• No new warnings
• New tests have been added
• All tests pass locally with my changes