This release is more scalable in memory usage

This release now allows (by default) change statistic computation without using lookup tables (either hash tables or arrays), which is much more scalable in terms of memory usage. This can also be more efficient in some cases even when large memory usage was not required.