Fixing opt-in for cluster tests and trust domain sync

Issues:
Fixes #456 and #460

Problem:
Developers must be able to opt-in to the cluster tests, since they
require a set of at least four bigip devices.

Also, a trust domain is implemented as a device group on the bigip
devices, and we must be able to sync that group when we add members to
the trust domain.

Analysis:
Used pytest.mark.skip to skip cluster tests unless a user has set a key
of 'run_cluster_tests' and a value of JSON/YAML truthiness.

The trust domain now instantiates a device group manager to sync to that
group when adding devices to the trust domain. This is a part of the
validation process in the device group and the trust domain.

Tests:
All unit and functional tests pass

Author: Paul Breaux

test/functional/cluster/test_cluster.py

@@ -144,32 +144,3 @@ def test_existing_failover_cluster(self, BigIPSetup):
 
         cm = ClusterManager(**kwargs)
         cm.teardown()
-
-    # The following tests were removed when scaling was taken out of
-    # the clustering feature.
-    def itest_scale_up_sync_failover(
-            self, BigIPSetup, ThreeBigIPTeardownSyncFailover
-    ):
-        a, b, c, d = BigIPSetup
-        bigip_list = [a, b]
-        cm = ClusterManager()
-        kwargs = {'devices': bigip_list,
-                  'device_group_name': DEVICE_GROUP_NAME,
-                  'device_group_partition': PARTITION,
-                  'device_group_type': 'sync-failover'}
-        cm.create(**kwargs)
-        cm.scale_up_by_one(c)
-
-    def itest_scale_up_down_up_down_sync_failover(
-            BigIPSetup, TwoBigIPTeardownSyncFailover):
-        a, b, c, d = BigIPSetup
-        bigip_list = [a, b]
-        cm = ClusterManager()
-        kwargs = {'devices': bigip_list,
-                  'device_group_name': DEVICE_GROUP_NAME,
-                  'device_group_partition': PARTITION,
-                  'device_group_type': 'sync-failover'}
-        cm.create(**kwargs)
-        for x in range(3):
-            cm.scale_up_by_one(c)
-            cm.scale_down_by_one(c)