[feature] AI API Gateway

[ivicac]

Summary

Introduce an AI API Gateway as a first-class automation capability: a unified entry point for routing, governing, and observing LLM traffic across multiple providers and models.

Capabilities

Configuration

• Providers (OpenAI, Anthropic, etc.) — CRUD
• Models registry — CRUD
• Routing policies — CRUD
• Budget configuration with enforcement modes
• Gateway settings

Observability

• Monitoring dashboard (requests, latency, errors, cost) via Recharts
• Workspace-scoped GraphQL operations for all gateway entities

Integration

• New sidebar entry in Automation (EE-gated)
• Liquibase schema migrations for gateway persistence

Acceptance Criteria

• Design spec merged and reviewed
• Backend module boots with migrations applied cleanly
• All CRUD pages (Providers, Models, Routing Policies, Budget, Settings) functional
• Monitoring dashboard renders live metrics
• EE edition gating verified (hidden in CE)
• Integration tests for gateway services