Iteration over links in flow interface

In addition to iteration over its children (atoms or subflows) each
pattern now provides iter_links() method that iterates over
dependency links. This allows engines to treat all patterns
in the same way instead knowing what structure each pattern expresses.

Change-Id: I52cb5b0b501eefc8eb56a9ef5303aeb318013e11

taskflow/flow.py

@@ -38,13 +38,6 @@ class Flow(object):
     a flow is just a 'structuring' concept this is typically a behavior that
     should not be worried about (as it is not visible to the user), but it is
     worth mentioning here.
-
-    Flows are expected to provide the following methods/properties:
-
-    - add
-    - __len__
-    - requires
-    - provides
     """
 
     def __init__(self, name, retry=None):
@@ -77,6 +70,21 @@ class Flow(object):
     def __len__(self):
         """Returns how many items are in this flow."""
 
+    @abc.abstractmethod
+    def __iter__(self):
+        """Iterates over the children of the flow."""
+
+    @abc.abstractmethod
+    def iter_links(self):
+        """Iterates over dependency links between children of the flow.
+
+        Iterates over 3-tuples ``(A, B, meta)``, where
+            * ``A`` is a child (atom or subflow) link starts from;
+            * ``B`` is a child (atom or subflow) link points to; it is
+              said that ``B`` depends on ``A`` or ``B`` requires ``A``;
+            * ``meta`` is link metadata, a dictionary.
+        """
+
     def __str__(self):
         lines = ["%s: %s" % (reflection.get_class_name(self), self.name)]
         lines.append("%s" % (len(self)))