DACPAC foundation: data-schema project + sprocs + CI/CD build & deploy

[rmartinsen-ucd]

Create the SDK-style DACPAC project (Microsoft.Build.Sql, default schema data) that owns all domain tables and processing sprocs, and wire it into the existing GitHub Actions pipeline following Walter's pattern.

Implementation

• Create the .sqlproj with a data schema and folders for tables/sprocs/functions.
• Build: build the .sqlproj and upload the .dacpac as an artifact.
• Deploy test (auto on push): sqlpackage /Action:Publish to the test Azure SQL DB using existing SQL_ADMIN_LOGIN/SQL_ADMIN_PASSWORD, with DropObjectsNotInSource=false.
• Deploy prod (gated): publish a /Action:Script artifact for review, then publish behind the manual deploy_prod / prod Environment gate.

Schema split (DACPAC owns data, EF owns app)

• DACPAC = data (domain tables + sprocs); EF = app (Run/RunStep/ImportLog control state). Schema prefix is the ownership token.
• EF confined to app (HasDefaultSchema("app"), app code reads data via Dapper); DACPAC confined to data (DropObjectsNotInSource=false). References point app → data only.
• Sprocs are parameter-in/results-out, never touch app, and are idempotent (app orchestrates exclusively).
• CI: fail build if EF migrations reference data. or the .sqlproj references app..

Related: #6 (tables), #7 (import).