TPOT doesnt save best pareto values in checkpoint folder

## Context of the issue

Currently when running through all the generations, the pareto "best values" gets updated as well as the best pipeline to date. However, I noticed with "periodic checkpoint folder" option that the best pareto values do not get saved over. I spent 4 days running TPOT and some of the best and simplest models - were not saved. For example, the gradient boosting model saved (with only one model) was 0.12 RMSLE but at the end of my pipeline run in the command line prompt, the RMSLE was 0.113.

## Expected result

All of the best pareto pipelines should be saved, not just the best pipeline to date

## Current result

Only the best pipeline to date is saved

## Possible fix

Also update the pareto pipelines - the best for each iteration