Update configuration in AWS so core and jobs can run healthily

[ships]

Issue(s) Resolved

The core container has not been able to healthcheck due to missing env vars, etc.

This PR sets everything in place so that it can run healthily (even if it doesn't do meaningful work yet).

Still TODO:

• diagnose why the frontend complains with something about supabaseUrl required -- seems to suggest that these supabase params are required in the next build step?

Test Plan

You can see the container running here

Screenshots (if applicable)

Optional

Notes/Context/Gotchas

Changes:

• Dockerfile now copies the generated prisma client directly, when building core.
• add external service secrets to AWS Secrets Manager
• .env file in docker context builds database URL using individual env vars
• jobs container gets the PG* vars that it needs instead of a .env file (these are the standard way, it's annoying that Prisma does not support this)
• adds a migrations "init container" that uses a container that has the whole monorepo
• RDS password generation no longer allows special characters like : -- these are technically legal for PG passwords, but require complicated handling in application code.
• once this is merged, we can manually dispatch a deploy from head of a PR branch if we want
• NEXT_PUBLIC_ vars are now handled by react-runtime-env.

Supporting Docs