Smac

Improvements
- Add an error when we get an empty dict data_to_scatter so that we can avoid an internal error caused in Dask precautiously
- Add experimental instruction for installing SMAC in Windows via a WSL.
- More detailed documentation regarding continuing runs.

Bugfixes
- Fix bug in the incumbent selection in the case that multi-fidelity is combined with multi-objective (1019).
- Fix callback order (1040).
- Handle configspace as dictionary in mlp and parego example.
- Adapt sgd loss to newest scikit-learn version.

Improvements
- Callbacks registration is now a public method of the optimizer and allows callbacks to be inserted at a specific position.
- Adapt developer install instructions to include pre-commit installation
- Add option to pass a dask client to the facade, e.g. enables running on a hpc cluster (983).
- Added scenario.use_default_config argument/attribute=False, that adds the user's configspace default configuration
as an additional_config to the inital design if set to True. This adds one additional configuration to the number of configs
originating from the initial design. Since n_trials is still respected, this results in one fewer BO steps
- Adapt developer install instructions to include pre-commit installation.
- Add option to pass a dask client to the facade, e.g. enables running on a hpc cluster (983).
- Add example for using a callback to log run metadata to a file (996).
- Move base callback and metadata callback files to own callback directory.
- Add a workaround to be able to pass a dataset via dask.scatter so that serialization/deserialization in Dask becomes much quicker (993).

Bugfixes
- The ISB-pair differences over the incumbent's configurations are computed correctly now (956).
- Adjust amount of configurations in different stages of hyperband brackets to conform to the original paper.
- Fix validation in smbo to use the seed in the scenario.
- Change order of callbacks, intensifier callback for incumbent selection is now the first callback.
- intensifier.get_state() will now check if the configurations contained in the queue is stored in the runhistory (997)

Improvements
- Clarify origin of configurations (908).
- Random forest with instances predicts the marginalized costs by using a C++ implementation in `pyrfr`, which is much faster (903).
- Add version to makefile to install correct test release version.
- Add option to disable logging by setting `logging_level=False`. (947)

Bugfixes
- Continue run when setting incumbent selection to highest budget when using Successive Halving (907).
- If integer features are used, they are automatically converted to strings.

Workflows
- Added workflow to update pre-commit versions (874).

Misc
- Added benchmarking procedure to compare to previous releases.

- Completely reimplemented the intensifiers (including Successive Halving and Hyperband): All intensifiers support multi-fidelity, multi-objective and multi-threading by nature now.
- Expected behaviour for ask-and-tell interface ensured (also for Successive Halving).
- Continuing a run is now fully supported.
- Added more examples.
- Updated documentation based on new implementation.
- Added benchmark to compare different versions.

Bugfixes
- Correct handling of integer hyperparameters in the initial design (531)

Bugfixes
- Fixed random weight (re-)generalization of multi-objective algorithms: Before the weights were generated for each call to ``build_matrix``, now we only re-generate them for every iteration.
- Optimization may get stuck because of deep copying an iterator for callback: We removed the configuration call from ``on_next_configurations_end``.

Minor
- Removed example badget in README.
- Added SMAC logo to README.

Big Changes
* We redesigned the scenario class completely. The scenario is implemented as a dataclass now and holds only environment variables (like limitations or save directory). Everything else was moved to the components directly.
* We removed runtime optimization completely (no adaptive capping or imputing anymore).
* We removed the command-line interface and restructured everything alongside. Since SMAC was building upon the command-line interface (especially in combination with the scenario), it was complicated to understand the behavior or find specific implementations. With the removal, we re-wrote everything in python and re-implemented the feature of using scripts as target functions.
* Introducing trials: Each config/seed/budget/instance calculation is a trial.
* The configuration chooser is integrated into the SMBO object now. Therefore, SMBO finally implements an ask-tell interface now.
* Facades are redesigned so that they accept instantiated components directly. If a component is not passed, a default component is used, which is specified for each facade individually in the form of static methods. You can use those static methods directly to adapt a component to your choice.
* A lot of API changes and renamings (e.g., RandomConfigurationChooser -> RandomDesign, Runhistory2EPM -> RunHistoryEncoder).
* Ambiguous variables are renamed and unified across files.
* Dependencies of modules are reduced drastically.
* We incorporated Pynisher 1.0, which ensures limitations cross-platform.
* We incorporated ConfigSpace 0.6, which simplified our examples.
* Examples and documentation are completely reworked. Examples use the new ConfigSpace, and the documentation is adapted to version 2.0.
* Transparent target function signatures: SMAC checks now explicitly if an argument is available (the required arguments are now specified in the intensifier). If there are more arguments that are not passed by SMAC, a warning is raised.
* Components implement a ``meta`` property now, all of which describe the initial state of SMAC. The facade collects all metadata and saves the initial state of the scenario.
* Improved multi-objective in general: RunHistory (in addition to RunHistoryEncoder) both incorporates the multi-objective algorithm. In other words, if the multi-objective algorithm changes the output, it directly affects the optimization process.
* Configspace is saved in json only
* StatusType is saved as integer and not as dict anymore
* We changed the behavior of continuing a run:
* SMAC automatically checks if a scenario was saved earlier. If there exists a scenario and the initial state is the same, SMAC automatically loads the previous data. However, continuing from that run is not possible yet.
* If there was a scenario earlier, but the initial state is different, then the user is asked to overwrite the run or to still continue the run although the state is different (Note that this only can happen if the name specified in the scenario is the same). Alternatively, an `old` to the old run is added (e.g., the name was test, it becomes test-old).
* The initial state of the SMAC run also specifies the name (if no name in the scenario is specified). If the user changes something in the code base or in the scenario, the name and, therefore, the save location automatically changes.

New Features
* Added a new termination feature: Use `terminate_cost_threshold` in the scenario to stop the optimization after a configuration was evaluated with a cost lower than the threshold.
* Callbacks are completely redesigned. Added callbacks to the facade are called in different positions in the Bayesian optimization loop.
* The multi-objective algorithm `MeanAggregationStrategy` supports objective weights now.
* RunHistory got more methods like ``get_incumbent`` or ``get_pareto_front``.

Fixes
* You ever noticed that the third configuration has no origin? It's fixed now.
* We fixed ParEGO (it updates every time training is performed now).

Optimization Changes
* Changed initial design behavior
* You can add additional configurations now.
* ``max_ratio`` will limit both ``n_configs`` and ``n_configs_per_hyperparameter`` but not additional configurations
* Reduced default ``max_ratio`` to 0.1.

Code Related
* Converted all unittests to pytests.
* Instances, seeds, and budgets can be set to none now. However, mixing none and non-none will throw an exception.