Josh Steadmon [off-list ref] writes:

Ah, sorry for misunderstanding. I've restored the test along with a
test_unconfig line for V3; however, doesn't this just turn into a "change
detector" test whose only purpose will be to fail if/when we change the
default value for this config option? 

Maybe a difference in opinion: "change detection" sounds like
non-essential changes to program behavior can cause a test to fail, but
default value handling seems quite essential to me. Not to mention that
default value handling is easy to break, so IMO guarding against
accidental regressions is pretty useful.

As for practical churn, I don't expect that we'll intentionally change
this default often.