Query returns incorrect [null] in one column, for all rows but one (dataset and query attached)

First of all, my apologies if I've overlooked something trivial or if this has already been reported. I'm a little shaky on proper terminology when it comes to "behind the curtain" of MariaDB so I may have missed something.

I attach here a mysqldump (problem_db.sql) of three tables, along with a query (query.sql) which is returning incorrect results.

(please note: the table stock_history is intentionally empty)

Expected result:

Returned result:

As you can see, the qty column is null for all but the first row, which is incorrect. It should have a value in all rows.

(As an aside, variations of this query/dataset can result in this bug returning null for all but the first row regardless of ORDER BY clause; the query/dataset presented here as a minimised case do not seem to do this)

I have found a number of actions that will each cause the bug to be hidden and the expected results to be returned:

• Modifying the WHERE clause in query.sql to "po_id >= 9936" will return the correct results for the rows presented above. I'm not sure if other rows are incorrect (you can always add LIMIT 7 to avoid seeing the other rows anyway)
• Deleting the index on po_detail.po_id
• Deleting the foreign key from po_queue.po_detail_id to po_detail.po_detail_id
• Removing the SUM aggregate and GROUP BY in the query's derived table "po_added"
• Removing the MAX aggregate and GROUP BY in the query's derived table "stock"
• Removing the derived table "stock" from the query

I have verified this bug on two systems (one under WSL) although both were using the same platform and MariaDB package.

Thanks in advance for your help!