Write evergreen parser to embed whole documents

[kylestratis]

This will also allow for integrating a built evergreen directly with a closely-linked existing evergreen.

[kylestratis]

When this is done, we will be able to add document metadata to the knowledgebase as well, and query against that