Use network 'heat-net' for integration_tests

Neutron support for ipv6 is not complete yet (ex. LBaaS
only supports ipv4). We'had some gate issues after dual
stack was enabled in devstack. We had put some tactical
fixes to resolve those[1][2].

This fix adds usage of heat-net/heat-subnet(only ipv4)
for integration tests. This also rolls back the earlier
tactical changes.

https://review.openstack.org/#/c/178576
https://review.openstack.org/#/c/178933/

Change-Id: Ia863edb8932b8dea5c4fa110c97dcfdadca85bb9
diff --git a/scenario/test_neutron_autoscaling.py b/scenario/test_neutron_autoscaling.py
index 9a1e634..e7aae19 100644
--- a/scenario/test_neutron_autoscaling.py
+++ b/scenario/test_neutron_autoscaling.py
@@ -40,7 +40,7 @@
             "image_id": self.conf.minimal_image_ref,
             "capacity": "1",
             "instance_type": self.conf.minimal_instance_type,
-            "fixed_subnet_name": self.conf.fixed_subnet_name,
+            "fixed_subnet": self.net['subnets'][0],
         }
 
         # Launch stack