sqllite experiment serialization

[octopuscabbage]

I think it would be useful to have sqllite experiment serialization. This would allow queries later on and potentially speedup saving/loading of large experiments (many thousands of experiments)

https://pypi.org/project/sqlitedict/

[octopuscabbage]

Probably the correct architecture here is to have a separate process that maintains database access and reads through a queue.