This PR adds an extra tests case that proves, what I consider, an unfortunate flaw in an optimization included a while ago ~around #188.
The explanation is that if we have a proof of absence and a proof of presence in the same leaf node, and that proof of absence is for a nil value, we get in a situation in which we can't figure out the appropriate stem.
This PR adds an extra tests case that proves, what I consider, an unfortunate flaw in an optimization included a while ago ~around #188.
The explanation is that if we have a proof of absence and a proof of presence in the same leaf node, and that proof of absence is for a
nilvalue, we get in a situation in which we can't figure out the appropriatestem.