There might be some places where delQ could/should be replaced with cleraIndex based on the intent of the call, but it shouldn't really be causing any correctness issues.
- if a test is doing a delQ to simulate an external user doing a delQ then that's a valid and correct usage.
- if a test is doing a delQ to "reset" test state to emulate a completely pristine solr collection, that's where cleraIndex is (now) a better choice – but it's not more correct
For 99% of all tests the diff is academic, but places where there is a diff are when tests muck with version numbers synthetically (ie: the original reason for this special syntax), have very specific assumptions about low level internal term stats (ie: the new updatable doc values tests, or perhaps some luke-esque tests) etc...
if you think there are tests that would be improved by switching from delQ to clearIndex in their test scafolding (ie: in Before/After methods, or when reseting some state) then sure – go ahead and open a new issue for those. But tests that do "normal" user requests, and do "normal" delq(matchalldocs) as part of that are just fine and certainly don't need .changed.
EDIT: after a few more minutes thought, added some more clarification about the correctness question above, and this followup comment ...
Personally: I don't know that it's worth the effort to go looking for places to make this change. My main concern was simply that if/when people write new tests, that may involve dependencies/assumptions on having a pristine index in each test method, having clearIndex work the way it does now is good, and will automatically save people headaches like the ones Ishan and I had recently.