I think we should discuss the design of the ServiceDB and if we really need the synchronization that is currently built into it. I feel like this is not required since levelDB is already safe to be used concurrently.

What do you think?

Synchronization for safe concurrent use is absolutely not the point of this design. The point is to decouple the service generating proofs from the consumers of these proofs (and round data). In this design, the Service simply produces round data and proofs. It has no knowledge about what happens to them afterward.

IMO it greatly simplifies testing as a consumer is now completely decoupled from the producer. They are connected with a channel only. Take a look at the unit tests. They don't need a mock for the hypothetical "database interface". It avoids creating untestable monsters like the atxBuilder in go-spacemesh where UTs are a mock hell.