We had to think carefully about some of these cases to come up with a correct and performant encoding scheme for columnar indexes. During this work we identified a list of interesting documents and expected results. This ticket tracks the work to convert that list into a jstest and commit that.