MySQL perfomance impact on COUNT DISTINCT when adding subquery to WHERE condition with 20000 rows in parent table

alternative_performant_having_query.sql

SELECT COUNT(DISTINCT a.angebot_id),
ash.angebot_status_datum,
(
    SELECT MAX(ash2.angebot_status_datum)
    FROM angebot_status_historie ash2
    WHERE ash2.angebot_id = a.angebot_id
) AS current_state
FROM
    angebot a
JOIN angebot_status_historie ash USING (angebot_id)
HAVING ash.angebot_status_datum = current_state
# duration 0,094 sec

original_performant_query.sql

SELECT COUNT(DISTINCT a.angebot_id)
FROM
    angebot a
JOIN angebot_status_historie ash USING (angebot_id)
# duration 0,062 sec

timeout_subquery_condtion_query.sql

SELECT COUNT(DISTINCT a.angebot_id)
FROM
    angebot a
JOIN angebot_status_historie ash USING (angebot_id)
WHERE ash.angebot_status_datum = (
    SELECT MAX(ash2.angebot_status_datum)
    FROM angebot_status_historie ash2
    WHERE ash2.angebot_id = a.angebot_id
)
# timeout

Alternativley I would use a HAVING clause and select the MAX() before. Unfortunately I am using a generator that has to reset the SELECT clause.

Is there a workaround?

Related to Doctrine and subqueries by @philipphoffmann:
http://www.philipphoffmann.de/blog/2012/08/29/a-bulletproof-pattern-for-creating-doctrine-subqueries-of-any-complexity/