more robust error handling.

[jangalinski]

When something fails "behind the scenes" (events can not be created because they are invalid, axon setup not working, ...) all tests become just "red", instead single steps failing.
It can be hard to debug/analyse what actually failed.

Typical exception: GIVEN/WHEN had issues, jgiven continues and has invalid state:

15:00:18.308 [main] ERROR c.t.jgiven.impl.ScenarioExecutor - The field resultValidator is required but has not been provided.
com.tngtech.jgiven.exception.JGivenMissingRequiredScenarioStateException: The field resultValidator is required but has not been provided.
	at com.tngtech.jgiven.impl.inject.ValueInjector.updateValues(ValueInjector.java:115)
	at com.tngtech.jgiven.impl.ScenarioExecutor.updateScenarioState(ScenarioExecutor.java:239)
	at com.tngtech.jgiven.impl.ScenarioExecutor.access$500(ScenarioExecutor.java:41)
	at com.tngtech.jgiven.impl.ScenarioExecutor$StageTransitionHandlerImpl.enterStage(ScenarioExecutor.java:172)
	at com.tngtech.jgiven.impl.intercept.StepInterceptorImpl.intercept(StepInterceptorImpl.java:80)
	at com.tngtech.jgiven.impl.intercept.ByteBuddyMethodInterceptor.interceptSuper(ByteBuddyMethodInterceptor.java:37)

Ideas:

• wrap the calls to axon-test/ResultValidator in try/catch
• instead of lateinit try to create usable scenarioStates/stateHolders so we can continue the test runs even if we can not really assert state.