[deploy/helm] Helm chart documentation: SQLite vs Postgres, ingress, autoscaling, common scenarios

[mfittko]

Summary

Add concise Helm chart documentation covering the common deployment scenarios without turning it into a giant “platform guide”.

This is explicitly not Kubernetes provisioning / CI/CD / GitOps.

Scope

• Document minimal install (single-replica, SQLite) with helm install example
• Document external PostgreSQL scenario (references to existing Secret for DATABASE_URL)
• Document external Redis scenario (values + env wiring)
• Document optional ingress + autoscaling usage (once implemented in [deploy/helm] Ingress + autoscaling hardening (HPA) for llm-proxy #207)
• Keep examples short; prefer copy-pastable helm install ... --set ... or small values.yaml snippets

Where

• Prefer a chart-local README (e.g. deploy/helm/llm-proxy/README.md) and/or a short pointer from docs/deployment/index.md.

Dependencies

• Depends on [deploy/helm] Secure secrets injection: MANAGEMENT_TOKEN + DATABASE_URL (external Postgres) #203/[deploy/helm] External Redis support (event bus + cache) and optional in-cluster Redis toggle #204/[deploy/helm] Ingress + autoscaling hardening (HPA) for llm-proxy #207 for the specific values keys.

Acceptance Criteria

• Users can deploy:
  • SQLite mode
  • External Postgres mode
  • External Redis mode
  • (When available) ingress + autoscaling

Parent: #49