Define and implement a basic "data" service for sinking project and run data into

[rtyler]

The object store can store artifacts and full logs, but there is still a need for a "data" service which can centralize active projects, and their relations to pipeline runs.

Whatever this is will need to be backed by a relational data store of some form. I'm thinking I can do something simple with sqlx and sqlite at the moment, but will need to provide a means to get into PostgreSQL or others later.