Refactor runner API to avoid creating testInstance (and other scenario-specific logic) in case of GPMC

[eupp]

This PR attempts to refactor the interaction of the Strategy class and the Runner class to make the Strategy runner-agnostic. The goal is to encapsulate the execution scenario specific logic related to data structures testing.

In particular, in this PR the testInstance object is not used (and thus not re-created) on each invocation when GPMC is used.

In more details, this PR:

• introduces ExecutionScenarioRunner specific to execution scenario running logic;
• introduces LambdaRunner specific to GPMC case when only lambda is used to run the test --- this runner avoids most complexities related to ExecutionScenarioRunner providing much simpler implementation;
• simplifies common Runner interface.

Unfortunately, this PR does not completely decouples ManagedStrategy from the execution scenario logic, there are some leftovers that should be handled in the future in separate PRs (as this one is already quite large):

• currentActorIsBlocking and concurrentActorCausesBlocking properties should probably be moved to ExecutionScenarioRunner;
• scenario object is used in several places in ManagedStrategy (related to Treat first actor (operation) as regular method call #452 (comment));
• number of threads in the scenario is used to distinguish between scenario threads and user-created threads --- in the future we should remove any distinction between the two in ManagedStrategy.