This is a list of ideas, each of which has a performance/accuracy trade off. We'll only pursue ideas that will provide more info about the feasibility of index-based CE, based on our initial findings from the perf/accuracy testing. We will re-run the perf/accuracy benchmarks to validate the below changes.

• Read into leaf pages for better accuracy for very small ranges (traversals end on the same leaf node). Could involve extending the API to indicate that the counts may be "exact"
• Read further into the tree for better accuracy on larger ranges. The initial implementation of WT-7408 stopped reading at the divergence node. We could consider reading one level further after divergence, or we could even get exact counts by computing the rank of both keys, then subtracting. This involves two full traversals of the index tree.