Make the AverageLearner only return new points ...

and introduce learner.pending_points to make it more similar to the rest of Adaptive.

I've added a test that will fail with the current master.

adaptive/learner/learner1D.py

@@ -147,22 +147,23 @@ class Learner1D(BaseLearner):
                 self.losses_combined[x_left, x_right] = loss
 
     def update_losses(self, x, real=True):
+        a, b = self.find_neighbors(x, self.neighbors_combined)
+        x_left, x_right = self.find_neighbors(x, self.neighbors)
+        losses_combined = self.losses_combined
+
         if real:
-            x_left, x_right = self.find_neighbors(x, self.neighbors)
             self.update_interpolated_loss_in_interval(x_left, x)
             self.update_interpolated_loss_in_interval(x, x_right)
             self.losses.pop((x_left, x_right), None)
             self.losses_combined.pop((x_left, x_right), None)
             a, b = self.find_neighbors(x, self.neighbors_combined)
             self.losses_combined.pop((a, b), None)
+
             if x_left is None and a is not None:
-                self.losses_combined[a, x] = float('inf')
+                losses_combined[a, x] = float('inf')
             if x_right is None and b is not None:
-                self.losses_combined[x, b] = float('inf')
+                losses_combined[x, b] = float('inf')
         else:
-            losses_combined = self.losses_combined
-            x_left, x_right = self.find_neighbors(x, self.neighbors)
-            a, b = self.find_neighbors(x, self.neighbors_combined)
             if x_left is not None and x_right is not None:
                 dx = x_right - x_left
                 loss = self.losses[x_left, x_right]
@@ -174,7 +175,7 @@ class Learner1D(BaseLearner):
                 if b is not None:
                     losses_combined[x, b] = float('inf')
 
-            losses_combined.pop((a, b), None)
+        losses_combined.pop((a, b), None)
 
     def find_neighbors(self, x, neighbors):
         if x in neighbors: