Make sure rabbitmq restarts on failures

Rabbitmq was not configured to restart on failures. It does not explain
why rabbitmq crashes (with exit code 255, see issue #80), but this should be sufficient
as a quick fix.

CHANGELOG.md

@@ -1,5 +1,6 @@
 # Changelog
 
+- 2018-10-30 [Bugfix] Fix rabbitmq restart policy
 - 2018-10-03 [Improvement/Bugfix] Fix and accelerate Android application build
 - 2018-10-02 [Improvement] Bump Open edX version to hawthorn.2
 - 2018-09-30 [Bugfix] Fix CMS celery worker, including export tasks

docker-compose.yml

@@ -33,7 +33,7 @@ services:
       memlock:
         soft: -1
         hard: -1
-    restart: "unless-stopped"
+    restart: unless-stopped
     volumes:
       - ./data/elasticsearch:/usr/share/elasticsearch/data
 
@@ -53,6 +53,7 @@ services:
     image: rabbitmq:3.6.10
     volumes:
       - ./data/rabbitmq:/var/lib/rabbitmq
+    restart: unless-stopped
 
   # Simple SMTP server
   smtp: