#667, move away from ansible timeout to env vars

docker_services' (of ansible) timeout didn't appear to effect the
timeout in use.  Setting COMPOSE_HTTP_TIMEOUT env variable for all
docker_services use was the most important in terms of getting a
successfull response.  Added DOCKER_CLIENT_TIMEOUT as a pair to that
based on the suggestion of
docker/compose#3927 (comment)

deploy/ansible/playbook.yml

@@ -28,6 +28,8 @@
     PCMT_TRAEFIK_CONF: "{{ pcmt_traefik_conf }}"
     PCMT_HOSTNAME: "{{ pcmt_hostname }}"
     PCMT_S3_BUCKET: "s3://{{ pcmt_hostname }}"
+    COMPOSE_HTTP_TIMEOUT: "120"
+    DOCKER_CLIENT_TIMEOUT: "120"
 
   pre_tasks:
     - include_tasks: tasks/python2-ubuntu.yml

deploy/ansible/tasks/pim.yml

@@ -35,7 +35,6 @@
     pull: no
     remove_orphans: yes
     state: absent
-    timeout: 120
   when: pcmt_profile == 'dev'
 - name: Prune pim volume for dev
   shell: docker volume rm -f app_pim
@@ -51,7 +50,6 @@
     pull: yes
     remove_orphans: yes
     state: present
-    timeout: 120
 - name: Restart fpm to wipe
   docker_service:
     project_src: /app