Fix report generation when multiple datasets are involved

[mariamedp]

• Fixed name in report generation files, was causing the reports to be overriden instead of creating one for each eval flow executed.
• Added small refactoring to reduce code duplications.

After the change, files for all eval flows executed are created. Eval flow name is used now in the file name.
Example with web classification use case:

[mariamedp]

This is the fix. evaluator.name is used for the file name instead of run_dataset.name.

This is because run_dataset is only a temporary variable that is used in a loop some lines above to iterate through all existing datasets. Hence, run_dataset here always has the same value since it's after the loop has run and has iterated through the whole list.