Do not sort dictionaries in assert_deepequal utility function

Sorting lists of dictionaries in assert_deepequal was causing inconsistencies
in unit test execution. To fix this, do not sort lists if their elements are
dictionaries.

https://fedorahosted.org/freeipa/ticket/3562

tests/test_xmlrpc/test_netgroup_plugin.py

@@ -464,12 +464,6 @@ class test_netgroup(Declarative):
                 truncated=False,
                 summary=u'2 netgroups matched',
                 result=[
-                    {
-                        'dn': fuzzy_netgroupdn,
-                        'cn': [netgroup2],
-                        'description': [u'Test netgroup 2'],
-                        'nisdomainname': [u'%s' % api.env.domain],
-                    },
                     {
                         'dn': fuzzy_netgroupdn,
                         'memberhost_host': (host1,),
@@ -478,6 +472,12 @@ class test_netgroup(Declarative):
                         'description': [u'Test netgroup 1'],
                         'nisdomainname': [u'%s' % api.env.domain],
                     },
+                    {
+                        'dn': fuzzy_netgroupdn,
+                        'cn': [netgroup2],
+                        'description': [u'Test netgroup 2'],
+                        'nisdomainname': [u'%s' % api.env.domain],
+                    },
                 ],
             ),
         ),

tests/util.py

@@ -316,6 +316,11 @@ def assert_deepequal(expected, got, doc='', stack=tuple()):
             raise AssertionError(
                 LEN % (doc, len(expected), len(got), expected, got, stack)
             )
+        # Sort list elements, unless they are dictionaries
+        if expected and isinstance(expected[0], dict):
+            s_got = got
+            s_expected = expected
+        else:
             s_got = sorted(got)
             s_expected = sorted(expected)
         for (i, e_sub) in enumerate(s_expected):