Support a json_full and yaml_full format for output

[nevinera]

(After #44)

Add two more formats, and another serializer, which can generate more thorough data, essentially a full dump of all the data structures constructed:

• changed-files (broken out by change source?)
• pipeline information including the parameters supplied, commands run, outcome objects
• executor information (varies by executor)
• options/config objects
• anything else you can think of.

The goal here is that, if someone has an issue, we can have them dump this file, and have all the data we need to understand what's going on, and probably to reconstruct the issue.

[nevinera]

Now that I think about it, non-open-source projects probably can't give us that information, since it'll have a bunch of details about source code in it -.-

I'm going to close this for now, it's a lot of work that might not be valuable after all.