Previously, The model was defined by terms from DNA and Protein Alphabet.
There were either "reversible" or "Non-reversible" models, where in the first: 	
 _Q[1][0] = _Q[0][1] * _freq[0] / _freq[1]; // stated that we find lossRate from gainRate and stationary freq.
 
 The freq value is also used as the Root freq at each likelihood computation,
 so using the "reversible" model alse means that the stationary freq = Root freq, where if lossRate was estimated independently
 in the "Non-reversible" - than the _freq is only addressed as Root freq.
 
 The models free parameters:
 M1 - gainRate, freq (both stat and root)
 M2 - gainRate, LossRate, freq (only root)
 
 In the royal society the model was re-written equivantly as:
 M1 - r_Q, freq (both stat and root) [where gainRate = r_Q*freq[1]] 
 M2 - r_Q, freq_Stat, and freq_Root [where freq_Stat was seperated since it was not used in _Q[1][0] = _Q[0][1] * _freq[0] / _freq[1];]
 

 