[N1QL] count(distinct field) gives different results for in operator with secondary index vs primary index

Test which fails:
./testrunner -i query2.ini -p doc-per-day=6 -t tuqquery.tuq_2i_index.QueriesIndexs.test_count_distinct

Query executed:
select count(distinct VMs[1].os) from default where VMs[1].os='windows' and tasks_points.task1 in [1,2,3]

Indexes present:
CREATE INDEX `idx` ON `default`(`job_title`) WHERE (`email` like "%@%.%")
CREATE INDEX `idx2` ON `default`(((`VMs`[1]).`os`),(`tasks_points`.`task1`))
CREATE INDEX `idx3` ON `default`(((`VMs`[1]).`os`),`name`) WHERE ((`tasks_points`.`task1`) = 1)

Result for query with secondary index:
[

]
Result for query with primary index:
[

]

Prathibha Bisarahalli:This seems to be related to MB-24284 though test in MB-24284 passes hence MB-24284 is verified as fixed.

Attaching dataset.