Evaluation stop criteria is not exact

[caranha]

Stop criteria is tested once every iteration. If we select the number of evaluations stop criteria, we may do more than the defined number of evaluations.

For example, suppose the stop criteria is 1000 function evaluations, and we have 300 subproblems. The program will stop at iteration 4, after 1200 evaluations have been performed.

For a very large total number of function evaluations, this may not matter much. But for more costly problems, it may become an issue.

Unfortunately, I'm not sure this is treatable with how we have structured the code and components: The two ways I know to deal with this is to truncate the last population (not possible if # of population == # of subproblems), or to bail out of the program, which may require us to set up some sort of observer in R.

Not urgent, but we will definitely want to tackle this problem in future versions of the package.