Optimize TBR to handle long sums

[ovdiiuv]

Currently when analyzing an expression for non linearity in TBR we visit the same expression a lot of times like in the example below, so it takes too much time to compile:

double f50x(double x){
  double a = 0;
  a = x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x+x;
  return a;
}

Same might relate to the Activity Analysis once it is merged.