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/average_learner.py

 # -*- coding: utf-8 -*-
 import itertools
 from math import sqrt
-import warnings
 
 import numpy as np
 
@@ -54,7 +53,7 @@ class AverageLearner(BaseLearner):
     def ask(self, n, add_data=True):
         points = list(range(self.n_requested, self.n_requested + n))
 
-        if any(True for p in points if p in self.data or p in self.pending_points):
+        if any(p in self.data or p in self.pending_points for p in points):
             # This means some of the points `< self.n_requested` do not exist.
             points = list(set(range(self.n_requested + n))
                           - set(self.data)
@@ -67,7 +66,7 @@ class AverageLearner(BaseLearner):
 
     def tell(self, n, value):
         if n in self.data:
-            warnings.warn(f"n={n} already exists in `learner.data` with value {self.data[n]}.")
+            # The point has already been added before.
             return
 
         if value is None:
@@ -110,7 +109,7 @@ class AverageLearner(BaseLearner):
 
     def remove_unfinished(self):
         """Remove uncomputed data from the learner."""
-        pass
+        self.pending_points = set()
 
     def plot(self):
         hv = ensure_holoviews()