Do not update average cost price if the quantity is negative

This is the same test done in Move._compute_product_cost_price. When the stock
quantity is below zero, the average cost price should not change.
And we process the incoming move first in order to keep quantity positive
where possible.

issue9484
review325801002

product.py

@@ -143,9 +143,7 @@ class Product(metaclass=PoolMeta):
             return move.from_location.type == 'storage'
 
         def compute_fifo_cost_price(quantity, date):
-            fifo_moves = self.get_fifo_move(
-                float(quantity),
-                date=current_moves[-1].effective_date)
+            fifo_moves = self.get_fifo_move(float(quantity), date=date)
 
             cost_price = Decimal(0)
             consumed_qty = 0
@@ -159,6 +157,11 @@ class Product(metaclass=PoolMeta):
             if consumed_qty:
                 return round_price(cost_price / Decimal(str(consumed_qty)))
 
+        # For each day, process the incoming moves first
+        # in order to keep quantity positive where possible
+        # We do not re-browse because we expect only small changes
+        moves = sorted(moves, key=lambda m: (
+                m.effective_date, out_move(m), m.id))
         current_moves = []
         current_out_qty = 0
         current_cost_price = cost_price
@@ -181,7 +184,7 @@ class Product(metaclass=PoolMeta):
                             m for m in out_moves
                             if m.cost_price != fifo_cost_price],
                         dict(cost_price=fifo_cost_price))
-                    if quantity:
+                    if quantity > 0 and quantity + current_out_qty >= 0:
                         cost_price = (
                             ((current_cost_price * (
                                         quantity + current_out_qty))