This PR is an overhaul of how the workbench integrates platypus-opt. This overhaul draws on the improvements made inside platypus. In passing, it also adds typing information, improves documentation, and ensures full test coverage.

Key new features

• adds support for starting the optimization from a user-provided initial population
• runtime convergence and storing of archives over time are handled automatically

Additional Improvements

• code simplifications
• bug fixes
• full test coverage
• typing information
• improved documentation

Done

1. Make use of new extensions functionality in Platypus for tracking progress, storing archives, and in GenerationalBorg
2. Simplify processing of platypus jobs to take advantage of the new Sample class. Substantial simplification of the code by adding helper methods to Sample (Sample._from_platypus_solution and Sample._to_platypus_solution), so we can easily go from platypus Solution to workbench Sample and from Sample to platypus Solution.
3. Minor bug fixes (always store archive if available, else store result)
4. Add a keyword argument for generator so we can start with a custom population --> now possible via initial_population keyword argument on BaseEvaluator.optimize
5. Ensure that runtime convergence information (operator probabilities/ e-progress) is always tracked and returned as part of the results.
6. unit tests
7. Perhaps split up optimization stuff over more files and add a subpackage inside em_framework to keep it organized and tractable.