Mako Range Read Testing Improvements

With a refactor of range reads in the works, an interesting case we want to test is a range read with a read-write heavy workload (as opposed to a read-only workload which is tested currently).

Currently our in memory ptree holds up to 5,000,000 versions (which is roughly the last 5 seconds of data). A part of a range read query involves combining the results read on disk (sqlite) with the ones held in memory (ptree). This case is tricky to test because reads that cross write boundaries need to happen within 5 seconds of each other, thus we cannot use random sets/clears and range reads (as we do currently in mako). Implementing this capability will allow us to test our average/worst case scenarios for range reads instead of simply the best case (no sets/clears)