Update unittest with run success check

[nkanazawa1989]

Summary

When you run Experiment.run this will suppress all errors incurred durning the execution and errors will be output as user warning

"Possibly incomplete analysis results: an analysis callback raised an error."

with stored error traceback. This is problematic behavior in unittest since CI might miss some unexpected bugs as maybe reported in #533.

This PR adds assertSuccess and assertFail checks that are expected to be called right after every run method call.

Details and comments

[nkanazawa1989]

I blocked this until #573 is resolved.

[nkanazawa1989]

Now this PR adds unittest for #599. Calling block_for_results method in unittest is replaced with self.assertComplete which also checks if experiment status becomes Done after execution otherwise raise an error.

[chriseclectic]

The added assert is basically functioning as a way to automatically block for results and check an experiment is Done, so I think the assert name should be changed to reflect that.
It is basically a more opaque that you are just checking:

assertEqual(data.block_for_results().status(), ExperimentStatus.DONE)

Also I think the assertFail test isn't needed and can just be removed, it would be better to be clear in the places where you are looking for errors to assert status = ERROR.

[nkanazawa1989]

Thanks Chris. These all make sense. I updated the PR according to your comments.