Add multi-service deployment stack for TRM platform

Initial pilot stack: redis + tcp-ingestion. Designed to grow as the
platform's other services land (processor, postgres+timescale, directus,
react-spa).

- compose.yaml: services as image: references with env-var-driven tags
  and ports. Redis is internal-only (no host port). TCP port 5027
  exposed for GPS device traffic. Persisted Redis volume.
- .env.example: documents TCP_INGESTION_TAG, INSTANCE_ID, PORT, LOG_LEVEL.
  Compose has defaults so the stack starts with no env config.
- README: Portainer Repository Stack instructions, manual deploy fallback,
  network model, planned-services list, why-separate-repo rationale.
- .gitignore: ignore .env

compose.yaml

@@ -0,0 +1,69 @@
+# TRM platform — deployment stack
+#
+# Deployed via Portainer Repository Stack:
+#   Repository: git.dev.microservices.al/trm/deploy
+#   Compose path: compose.yaml
+#   Branch: main
+#
+# Images are built and pushed by each service's own Gitea workflow.
+# This file references them by tag and runs them as a coordinated stack.
+#
+# Before first deploy on the host: `docker login git.dev.microservices.al`
+# (Portainer can store registry credentials in its UI; configure once.)
+#
+# Environment variables are populated from Portainer's stack environment
+# config (or a `.env` file alongside this compose for non-Portainer hosts).
+# Defaults are provided via `${VAR:-default}` so the stack starts with no
+# explicit configuration on a fresh deploy.
+
+name: trm
+
+services:
+  # -------------------------------------------------------------------
+  # Redis — telemetry queue + (future) connection registry for Phase 2.
+  # Internal-only; no host port mapping.
+  # -------------------------------------------------------------------
+  redis:
+    image: redis:7-alpine
+    expose:
+      - '6379'
+    volumes:
+      - redis-data:/data
+    restart: unless-stopped
+    healthcheck:
+      test: ['CMD', 'redis-cli', 'ping']
+      interval: 10s
+      timeout: 3s
+      retries: 5
+
+  # -------------------------------------------------------------------
+  # tcp-ingestion — Teltonika telemetry TCP server.
+  # Built by git.dev.microservices.al/trm/tcp-ingestion's Gitea workflow.
+  # -------------------------------------------------------------------
+  tcp-ingestion:
+    image: git.dev.microservices.al/trm/tcp-ingestion:${TCP_INGESTION_TAG:-main}
+    depends_on:
+      redis:
+        condition: service_healthy
+    ports:
+      # Devices connect to this port. Use `${HOST_BIND_IP:-0.0.0.0}:5027:5027`
+      # if you want to restrict which host interface accepts connections.
+      - '${TCP_INGESTION_PORT:-5027}:5027'
+    environment:
+      NODE_ENV: production
+      INSTANCE_ID: ${TCP_INGESTION_INSTANCE_ID:-stage-1}
+      REDIS_URL: redis://redis:6379
+      LOG_LEVEL: ${LOG_LEVEL:-info}
+    restart: unless-stopped
+
+  # -------------------------------------------------------------------
+  # Future services land here:
+  #   - processor:        consumes telemetry from Redis, writes to DB
+  #   - postgres:         PostgreSQL with TimescaleDB extension
+  #   - directus:         business-plane API + admin UI
+  #   - react-spa:        front-end (static, served via nginx or Caddy)
+  # See ../docs/wiki/ for the platform architecture.
+  # -------------------------------------------------------------------
+
+volumes:
+  redis-data: