fix: align quickstart with d0a1.es ecosystem and current reality

Hardening pass over the quickstart repo to make it match the actual
d0a1.es stack and remove long-standing drift.

- README: align with 8 GB RAM reality across all sections, add
  explicit security warning on WEBUI_AUTH=false, remove dead link
  to d0a1.es/vps, drop stack services not part of d0a1.es
  (n8n, WordPress, Qdrant). Re-point the d0a1 section to the
  services that actually run at git/mail/app.d0a1.es.
- docker-compose: move all tunables (ports, default model, auth,
  secret key, keep-alive) to .env with sensible defaults. Add
  commented GPU passthrough block for NVIDIA hosts.
- deploy.sh: pre-flight checks (docker, compose, daemon),
  copy .env.example on first run, auto-detect system RAM on
  Linux and macOS and warn on sub-8 GB hosts.
- Add .env.example documenting every variable and required RAM
  per model.
- Add MIT LICENSE (was promised in README but missing).
- Add .gitignore (local .env, editor cruft, OS junk).
- .gitattributes: use a single rule (text=auto eol=lf) to cover
  every text file, not just *.sh - fixes CRLF warnings on new
  text files on Windows checkouts.

Refs: d0a1/quickstart rename (companion: new repo at
git.d0a1.es/d0a1/quickstart).

docker-compose.yml

@@ -1,56 +1,39 @@
-# d0a1 quickstart — IA agéntica local, sin coste, sin VPS
-# Ejecuta: ./deploy.sh
-# Abre: http://localhost:3000
-#
-# Requisitos: Docker + 8 GB RAM mínimo
-# Incluye: Ollama (LLM local) + Open WebUI (interfaz web)
-# Modelo por defecto: llama3.2:3b (~2 GB, corre en 8 GB RAM)
 services:
   ollama:
+    image: ollama/ollama:latest
     container_name: d0a1-ollama
-    image: ollama/ollama:0.30.8
     restart: unless-stopped
     ports:
-      - "127.0.0.1:11434:11434"
+      - "${OLLAMA_PORT:-11434}:11434"
     volumes:
       - ollama_data:/root/.ollama
     environment:
-      OLLAMA_HOST: "0.0.0.0"
-      OLLAMA_KEEP_ALIVE: "5m"
-      OLLAMA_NUM_PARALLEL: "1"
-    healthcheck:
-      test: ["CMD", "ollama", "ps"]
-      interval: 15s
-      timeout: 5s
-      retries: 10
-      start_period: 30s
+      - OLLAMA_KEEP_ALIVE=${OLLAMA_KEEP_ALIVE:-5m}
+    # GPU passthrough: descomenta la sección deploy si tienes GPU NVIDIA
+    # deploy:
+    #   resources:
+    #     reservations:
+    #       devices:
+    #         - driver: nvidia
+    #           count: all
+    #           capabilities: [gpu]
 
-  openwebui:
-    container_name: d0a1-openwebui
-    image: ghcr.io/open-webui/open-webui:0.9.6
+  open-webui:
+    image: ghcr.io/open-webui/open-webui:main
+    container_name: d0a1-open-webui
     restart: unless-stopped
     ports:
-      - "3000:8080"
+      - "${WEBUI_PORT:-3000}:8080"
     volumes:
-      - openwebui_data:/app/backend/data
+      - open_webui_data:/app/backend/data
     environment:
-      OLLAMA_BASE_URL: "http://ollama:11434"
-      WEBUI_NAME: "d0a1 quickstart"
-      WEBUI_AUTH: "false"
-      DEFAULT_LOCALE: "es"
-      ENABLE_SIGNUP: "true"
-      ENABLE_WEB_SEARCH: "false"
-      ENABLE_DIRECT_CONNECTIONS: "false"
+      - OLLAMA_BASE_URL=http://ollama:11434
+      - WEBUI_AUTH=${WEBUI_AUTH:-true}
+      - WEBUI_SECRET_KEY=${WEBUI_SECRET_KEY:-}
+      - DEFAULT_MODELS=${DEFAULT_MODEL:-llama3.2:3b}
     depends_on:
-      ollama:
-        condition: service_healthy
-    deploy:
-      resources:
-        limits:
-          memory: 2G
+      - ollama
 
 volumes:
   ollama_data:
-    driver: local
-  openwebui_data:
-    driver: local
+  open_webui_data: