Don't fail the parser on invalid keys
I like the concept of having separate methods for validating and parsing RAML files. However there is at least one case where the parser also validates: If there is an unknown key in the RAML file, the parser throws an exception. It shouldn't do that. Instead, it should just skip the key and continue parsing the rest of the file. In this way it would be possible to at least read the valid parts of an invalid RAML file. We have an extension scenario in which we use RAML files that are only partially valid.
I'll provide a patch with a fix and a test case that illustrates the problem.
Aha! Link: https://mulesoft-roadmap.aha.io/features/APIRAML-139
Fixing this issue would also resolve issue #82