Added "Maximum Likelihood for Y=F(X;A)" in "Function Fit".

For the regression problem: Y=F(X；A) + error with the probability (density) function P(X;A) for Y is given, parameters, A1, A2 , … are estimated by maximum likelihood method.

By using Cauchy distribution, or more generally Pearson family of distributions, or Huber function for P(X;A), this can be used for robust regression. For details, see the Pages 3 and 4 in the sample file "Robust.kyp".