Fluidml

What's Changed
* Bug fix for tmux logging
* Enable nested multiprocessing by setting workers as non-daemonic
* Adds function to convert legacy local file storage structure (0.2.X) to new format
* Mp bug fixes by LarsHill in https://github.com/fluidml/fluidml/pull/49


**Full Changelog**: https://github.com/fluidml/fluidml/compare/v0.3.2...v0.3.3

What's Changed
* Fixes different `str` Mixin for `Enum` behavior in python 3.11.
* Enum fix python 311 by LarsHill in https://github.com/fluidml/fluidml/pull/48


**Full Changelog**: https://github.com/fluidml/fluidml/compare/v0.3.1...v0.3.2

Fixes an unintended `ImportError` caused by a missing `mongoengine` package, which should not occur due to `mongoengine` being optional.

This release contains a major API update to FluidML and introduces breaking changes compared to version 0.2.0.
We refer to the README and documentation for further details.

The main updates are:

- Cleaner API with less boilerplate. E.g. does not expose Swarm anymore and removes "publishes" and "expects"
- Advanced grid search expansion functionality
- Complete refactoring of Task Execution logic (with/without multiprocessing)
- Addition of TaskStates, "running", "finished", "killed", etc.
- Option for Tmux logging in multiprocessing. Each worker logs to a dedicated pane in a new tmux session (automatically created)
- Includes lazy loading for the Results Store
- Introduces Sphinx API documentation (to be further improved)
- Automatic run name creation, if no run name is provided.
- Minimum Python version increased to 3.7
- Run CI tests for 3.7 - 3.11
- Adds pre-commit-hooks (e.g. black, isort) to the repo
- Better error handling and adds the option to continue unaffected tasks if an error occurs in the pipeline.

- Added console and bokeh graph visualization
- Added delete method to result store
- Fixed race condition problem with locks in result store
- Added parameter coupling for grid search expansion method "product"
- Implicitly infer expects list for successor task

- Fixed a bug where the task graph isn't created when only 1 task is instantiated.
- Decoupled the optional install of rich and tblib.
- Added the option to select the gs_expansion_method, 'product' (default) or 'zip'.
- Added better documentation for reduce=True in README and examples.
- Added documentation for gs_expansion_method.