Deployment plan

[josephjclark]

Here's what I think we need to do:

• On merge to main, build, push and deploy straight to apollo-staging.open.org
• Enable manual deploy of staging to apollo.open.org
• Have some kind of end to end tests on the staging server (against key lightning services, not AI stuff)

Lightning will (eventually) depend on the production server for metadata and adaptor docs. So we want to minimise the risk of accidentally breaking that. The staging split with 2e2 tests would help.

We don't need to do all this on day one - the tests and prod server are only needed when lightning starts migrating services.

The CLI should probably always speak to the staging server by default - so that if you publish a change you can more easily test it through the CLI.

The CLI should have --local --staging and --prod flags to all apollo commands