[EPIC] Port ROR Reconciler to Python Application

[adambuttrick]

Summary

We should port ROR Reconciler to be a lightweight Python application. The existing service is a ~190-line Sinatra app running inside a Phusion Passenger/Nginx Docker container (phusion/passenger-ruby32:3.1.6). This should be replaced with a minimal Python web framework (e.g., Flask or FastAPI) to align the service with the use of Python across the rest of the ROR's services.

Tasks

• Rewrite all endpoints in Python with equivalent behavior:
  • GET / — redirect to ror.org
  • GET /heartbeat — health check
  • GET /reconcile — service metadata
  • POST /reconcile` — reconciliation queries (batch support)
  • GET /suggest — autocomplete suggestions
  • GET /flyout — inline preview data
  • GET /preview/* — HTML preview of ROR records
• Maintain W3C Reconciliation API compliance
• CORS support
• Create a new Dockerfile with minimal base image
• Update docker-compose.yml for local development
• Update GitHub Actions workflows (dev, staging, prod)
• Update Terraform variables
• Improved test suite
• CloudWatch monitoring for container lifecycle

Acceptance criteria

• All existing endpoints return equivalent responses to the Ruby implementation
• OpenRefine reconciliation workflows function identically against the Python service
• Test suite covers all endpoints
• CI/CD pipelines (dev/staging/prod) build and deploy the Python image
• CloudWatch alerts are in place for container restart anomalies
• No degradation in response times under normal load
• Documentation (README) updated to reflect the new stack