Improvements for multi-tenancy

[robfromboulder]

• 'viewzoo' is hardcoded in JdbcClient (can't use mutliple instances)
• expand docs on using multiple zoos (multiple connectors, filesystem & JDBC together)
• document use for tenancy (one zoo per tenant to map data sources and configurations)
• quantify how many views can be created? (create performance test for this)

[huw0]

I wondered if you could expand more on your plans for multi-tenancy.

The git-backed views mentioned in the project description is exactly what we do. However it is also handy to have a sandbox where a user can test views prior to raising a PR. Ideally this should be private to the user so that these aren't used by others before going through the PR process.

This can kind-of be implemented already with a Trino custom authentication layer that allows users to create schemas based on their username. But it'd be nice not to have to do this and instead have a 'sandbox' schema that actually maps to a separate folder/tenant for each user.

Would this be something you'd consider in-scope for multi-tenancy?

[robfromboulder]

I didn't have what you're describing specifically in mind here, but I'd be excited to explore this further.