MyISAM/Aria descending index compression is not as efficient as ascending

There was some effort made in 10.7 to store UUID efficiently (MDEV-26664).
Apparently the logic does not apply (or doesn't apply well enough) to descending indexes which are being introduced in 10.8 in the scope of MDEV-13756.

Below is based on type_uuid.type_uuid_myisam test.

CREATE TABLE t1 (a UUID, KEY(a)) ENGINE=MyISAM;

--delimiter $

FOR i IN 0..8191 DO

INSERT INTO t1 VALUES (UUID());

END FOR $

--delimiter ;

SELECT

INDEX_LENGTH, DATA_LENGTH, INDEX_LENGTH/DATA_LENGTH

FROM INFORMATION_SCHEMA.TABLES

WHERE TABLE_SCHEMA='test' AND TABLE_NAME='t1';

DROP TABLE t1;

With a standard ascending key it gives the result

INDEX_LENGTH	DATA_LENGTH	INDEX_LENGTH/DATA_LENGTH

79872	139264	0.5735

With a descending index (KEY(a DESC)) it produces

INDEX_LENGTH	DATA_LENGTH	INDEX_LENGTH/DATA_LENGTH

150528	139264	1.0809

Which, according to the same test's criteria, is a middle ground between "packed" and "not packed".