Automate Post-Install for Windmill

[IamJeffG]

Feature Request

https://github.com/ConservationMetrics/gc-forge/issues/104#issuecomment-5036356206 shows that 18% of our time in a new deployment goes toward Post-Install for Windmill!

TODO:

• Understand what has to happen there and what, if anything, we may want to automate e.g. with a script.
• Maybe instead of a script, can we push things upstream somehow?
• Consider also which pieces are better to have a user do, for learning reasons (doesn't mean we wouldn't still script it for internal deploys by experts)

User Stories & Acceptance Criteria

UI Sketch (or Screenshot)

Implementation Plan

(may be added later, before work begins)

[rudokemper]

I think there is a lot that we can do here. My quick mental model of what is currently done manually:

1. First-time login with admin and changeme (defaults)
2. Set basic settings like turning Telemetry off, providing auth0 domain
3. First-time sync of Windmill Hub resources (which itself is a Windmill script!)
4. Create gc-<alias> workspace
5. Register Windmill workspace with wmill, and run wmill sync push from GC Scripts Hub
6. Next, set up a bunch of resources that we provide at the outset like...
   • CoMapeo archive server
   • auth0 m2m client for metrics
   • postgres db connection
   • Mapbox account
   • GFW access
   • Local Contexts access
   • twilio and GCP, in case we will provide alerts
7. Configure Operator permissions and give Groups access to f/ directories
8. Schedule the GC Metrics script

I think the Windmill client basically dogfoods its own API for handling much of this stuff, and can therefore be automated.

Bonus points if we somehow use stack.yaml seeding (c.f. #157) to pre-populate some stuff like Postgres DB connection.

GFW, Local Contexts, GCP and Twilio are examples of things that don't change from partner to partner. We can seed those values from secrets kept somewhere.

Some things like getting a CoMapeo archive server API key is trickier as that is generated and set as an env var on app initialization - we don't set it in stack deploy. Although we could. Actually now that I think about it, it's mainly the CoMapeo archive server that is like this.