Revision

[murrayds]

Basic revisions of the paper following feedback from science advances. Mainly, I attempted to do the following:

• de-emphasize the "prediction" and "performance" aspects of the work. Instead, make it very clear that we are attempting to "validate" the embedding
• Draw more attention to the unique topology of the emebdding space, and the unique opportunities this brings in terms of methodology
• Other minor changes to the language throughout

I also updated the yml and stuff a bit