Helm charts + MicroK8s deployment path

[BrewingCoder]

Summary

Create a Helm chart for HoldFast and document a MicroK8s-based deployment path as the canonical production deployment method. Docker Compose remains for quick local dev.

Motivation

Docker Compose is easy but doesn't exercise real k8s deployment concerns. A 3-node MicroK8s cluster on Proxmox gives a realistic prod-like environment for development and integration testing.

Scope

• Helm chart covering all HoldFast services:
  • PostgreSQL (or dependency on upstream chart)
  • ClickHouse
  • Redpanda (Kafka-compatible, single container)
  • Redis
  • OTeL Collector
  • Backend (.NET)
  • Frontend
• values.yaml with sensible defaults for self-hosted deployment
• MicroK8s setup runbook (3-node: 1 control plane + 2 workers)
• CI pipeline to lint/validate the chart on PRs
• docs/DEPLOYMENT-K8S.md covering install, upgrade, and config

Out of scope

• HA for stateful services (Postgres, ClickHouse) — single replica for now
• Cloud-managed k8s (EKS, GKE, AKS) — MicroK8s / bare metal only

Dependencies

• .NET 10 backend replacement — Phase 0+1 scaffold #56 (.NET backend cutover) should be merged first so the chart targets the .NET image

[github-actions]

AI Triage: Create Helm charts and MicroK8s deployment documentation as the canonical production deployment method for HoldFast.

Labels applied: area:infra, enhancement, priority:high, component:docker