Handling "Continue on Failure"

Right now, if you you have a step that has continue on failure and the step fails in a run, the output of it will be just empty, users can’t manually add handling logic depending on the output which is blocking the full experience.

We already have a pr open, the pr will do two things:

1) If the step is tested and it fails, the output is saved so users can use the output in their error handling logic.

2) If the step output relies on the error, if the error message is a JSON we parse it, otherwise we keep it a string.
☝️ this cannot be played with in the future so we really have to think about it because a lot of people logic can be built on this simple realization.