Use from positional import positional syntax

This is always how i expected the class to be used and is why the WARN
and EXCEPT variables are on both the module and the class. It's much
neater and in the style of from pprint import pprint.

The tests should reflect the way we expect people to consume it.

positional/tests/test_positional.py

@@ -15,22 +15,22 @@ import logging
 import six
 import testtools
 
-import positional
+from positional import positional
 
 
 class TestPositional(testtools.TestCase):
 
-    @positional.positional(1)
+    @positional(1)
     def no_vars(self):
         # positional doesn't enforce anything here
         return True
 
-    @positional.positional(3, positional.positional.EXCEPT)
+    @positional(3, positional.EXCEPT)
     def mixed_except(self, arg, kwarg1=None, kwarg2=None):
         # self, arg, and kwarg1 may be passed positionally
         return (arg, kwarg1, kwarg2)
 
-    @positional.positional(3, positional.positional.WARN)
+    @positional(3, positional.WARN)
     def mixed_warn(self, arg, kwarg1=None, kwarg2=None):
         # self, arg, and kwarg1 may be passed positionally, only a warning
         # is emitted
@@ -62,7 +62,7 @@ class TestPositional(testtools.TestCase):
 
         self.assertIn('takes at most 3 positional', logger_message.getvalue())
 
-    @positional.positional(enforcement=positional.positional.EXCEPT)
+    @positional(enforcement=positional.EXCEPT)
     def inspect_func(self, arg, kwarg=None):
         return (arg, kwarg)
 
@@ -72,11 +72,11 @@ class TestPositional(testtools.TestCase):
         self.assertRaises(TypeError, self.inspect_func)
         self.assertRaises(TypeError, self.inspect_func, 1, 2)
 
-    @positional.positional.classmethod(1)
+    @positional.classmethod(1)
     def class_method(cls, a, b):
         return (cls, a, b)
 
-    @positional.positional.method(1)
+    @positional.method(1)
     def normal_method(self, a, b):
         self.assertIsInstance(self, TestPositional)
         return (self, a, b)