Hi Brenton,
(1) The Blueprints-compatible FDB graph layer that you linked was more of a prototype/demo than a production-quality layer. IIRC, Mike McMahon was working on a FoundationDB storage engine for TitanDB at one point, and I believe got it to a higher level of maturity. I have no idea whether Titan is still a popular database or not, but either way I don’t think it would be too hard to write a proper graph layer for FDB. I’d be happy to help and/or advise on any such effort.
(2) Your question about indexes, and Alec’s thorough but slightly daunting answer, is one reason I’d like to resurrect the document layer (as discussed here). It had a pretty fast and sophisticated implementation of the procedure Alec describes, and one mostly decoupled from the details of Mongo compatibility, and even from the document data model (IIRC, the indexing system only cared about hierarchical “field” or “paths”, and had a pluggable type system).