The promql testing framework already allows specifying a counter reset hint in a native histogram that is part of the expected result. However, the counter reset hint is then not compared with the one in the actual result.

Example of a test specifying a counter reset hint (which isn't really tested):

This is probably not as easy as adding this comparison. Ideally, we could implement a semantics where not specifying the counter reset hint means "I don't care about the counter reset", while an explicitly specified counter reset hint means "please check this". Otherwise, we have to rework all tests to explicitly specify the hint, as long as it is not the default (which is counter_reset_hint:unknown).