68: Support for dry-run to print undefined steps

[TitouanVanBelle]

This pull request allow users to use the dry run feature described in #68 the following way

[Cucumberish instance].dryRun = YES;
[Cucumberish instance].dryRunLanguage = CCILanguageObjectiveC;

The results are printed in the logs the following way

dryRunLanguage = CCILanguageObjectiveC

2017-06-16 16:52:22.892 xctest[39431:10863180] 

Dry Run Results
=====================================================
Found the following undefined steps:

Given(@"^a Given statement$", ^(NSArray<NSString *> *args, NSDictionary *userInfo) {

});

And(@"^an And statement$", ^(NSArray<NSString *> *args, NSDictionary *userInfo) {

});

=====================================================

dryRunLanguage = CCILanguageSwift

2017-06-16 16:52:22.892 xctest[39431:10863180] 

Dry Run Results
=====================================================
Found the following undefined steps:

Given("^a Given statement$") { (args, userInfo) in

}

And("^an And statement$") { (args, userInfo) in

}

=====================================================

When all steps are defined, the following is printed

Dry Run Results
=====================================================

All the steps are defined

=====================================================

[brentleyjones]

Initial review looks good. Do you think you could add some tests for this, so we don't regress on the behavior later (since not everyone will use this feature)?

[TitouanVanBelle]

Sure. Any idea what kind of tests would fit better for this feature? Do you think it makes sense to add tests to the CucumberishTests target for this?

[brentleyjones]

Yeah, add tests to CucumberishTests target. After that I'll get to this after the 4th of July. Busy at work and then a small vacation.

[TitouanVanBelle]

I added some test to the CucumberishFeatureDefinition target because the step definitions form the CucumberTests target are using @"(.*)" which makes it hard to have undefined steps.

I had to add a log manager to assert that the logs were properly printed. It might a bit overcomplicated for the task but it does the job.

I also had some problem since the steps are not executed when dryRun is enabled so I had to do the assertion in the afterFinish block. The problem with this solution is that it will fail the last scenario executed if the assert fails. So the dryRun scenario should always be executed last and that's why I added numbers to the feature names.

[brentleyjones]

Typo: tep

[brentleyjones]

Thanks @TitouanVanBelle!