[EM] Fine version of bias-test (small, brief modification)

[Michael Ossipoff]

Instead of seeking to make the ratio of the average s/q of the small states 
and large states as close to 1 as possible, the fine version of my bias test 
seeks to make the Pearson correlation of the average q and the average s/q 
of the cycles as close to zero as possible.

Cycles are a finer division. Since rounding is done in the cycles, one 
wouldn't want a finer division. Because of s/q disparity resulting from 
states being in different parts of their cycles, the cycle is the finest 
division for which it's meaningful to talk about equal s/q.

Mike Ossipoff