[FIX] purchase_stock: clean non deterministic test

test_reordering_rule_1 try to ressuply 10 from RR.
However since 15.0 the picking `action_confirm` will
launch himself the required orderpoints to order the
missing quantity. So the run scheduler is useless but
should not order the quantity again since it's already
process.

But in the test the cache for the orderpoint is not
always consistent with the data. It happens that the
quantity forecast remains -10 even after the RFQ creation.

It happens due to the savepoint. It will flush the data
and take a random env. However in the enviroment, the
active_test key could be there. And it will recompute
all the fields. So it goes through `_compute_rules` that
call `search_rule`. There is an archived route from supplier
to stock. So it will pick this rule instead of the buy route.
`_compute_lead_days` will not have the supplier delay as
a result and a wrong forecasted date. `_compute_quantities`
on product will check the virtual available for this date and
will only have the out move and not the in since the domain
is incorrect with the wrong lead days date...

I can't do much since it's due to the orm.
- Remove the savepoint but it will require a global refactoring
- Fix the ORM, it's planned but difficult.

So for now I just remove the context key form the context in the
compute.

Note: this could happens outside the test when there is more than
1000 RR.

closes odoo/odoo#121106

Signed-off-by: Tiffany Chang <tic@odoo.com>

addons/stock/models/product.py

@@ -566,7 +566,7 @@ class Product(models.Model):
         warehouse = location.warehouse_id
         if not warehouse and seen_rules:
             warehouse = seen_rules[-1].propagate_warehouse_id
-        rule = self.env['procurement.group']._get_rule(self, location, {
+        rule = self.env['procurement.group'].with_context(active_test=True)._get_rule(self, location, {
             'route_ids': route_ids,
             'warehouse_id': warehouse,
         })