BUG: Fix div by 0 error due to changed return type.

When calculate_positions_value used np.dot, the return type was a
np.float64. Which allows the use of 0.0 in division to not raise an
exception.

Fix by expliciting creating an np.float64 with 0 value.

zipline/finance/performance/period.py

@@ -351,7 +351,7 @@ class PerformancePeriod(object):
 
     def calculate_positions_value(self):
         if len(self.position_values) == 0:
-            return 0
+            return np.float64(0)
 
         return sum(self.position_values)