Horizontal scaling

Ideally, we want to allow larger organizations to scale instances horizontally, which access the same backing data.

We should strongly consider switching to a concurrent database of some sort.

The biggest issue is that the samod/subduction instances wouldn't be able to talk to each other and inform of writes.

No files or tests are named. Start by tracing how samod/subduction instances access backing data and how writes are communicated, then define the requirements for multiple instances to share data and coordinate writes. Done means the horizontal-scaling design is documented and validated against those coordination requirements.

Written by the indexing model from the issue text.