Added Smart System for quick-wide Experiments

[ozguraslank]

Explanation

• If user goes with 'quick' and then 'wide' without changing validation params, FlexML will only append non-existed models to save time
• random_state's default value is set to 42 instead of None, but user will be able to set it to None. If It's None, every experiment's result will be different since data will be prepared with different random seed in each execution
• Changed naming of the variables starts with '_current' since they are not clear, set them to start with '_last' since they keep the information of the last experiment's info
• Adjusted model performance table on jupyter notebook. It won't highlight any values If there is only one result in the table since you can't compare a single result