Pycomex

-------------------

Added a *Plugin System* to the pycomex framework. This is a major new feature which allows to extend the
functionality of the framework in a modular way. The plugin system is custom and implemented via hooks that
are accessible through a global singleton config instance.

- Added the ``pycomex.plugin`` module which contains the ``Plugin`` class and the ``PluginManager`` class
- Added the ``pycomex.config`` decorator which can be used to define hooks in the plugin modules
- Added the "weights_biases" plugin which is a simple example of how to use the plugin system. It implements
a weights and biases integration for the pycomex experimentation framework. In addition to the local artifact
folders it is now also possible to log the resuls to the online dashboard of weights and biases.
- Added some unittests for the config and plugin system

-------------------

- fixed a crucial bug that would break the experiment modules if no module level doc string exists
- Added the ``get_experiment`` method which allows to easily get the Experiment object instance based
on a given absolute experiment module path.
- Added the example ``07_meta_experiments.py``

-------------------

- Fixed a breaking bug during the construction of Experiment instances
- Added information about possible hooks to the ``info`` cli command as well

-------------------

Added the "testing" functionality as it's own feature to the Experiment class

- it is now possible to define the hook with the necessary code to put the experiment into testing mode using
the ``Experiment.testing`` function and the ``__TESTING__`` magic parameter.
- Added a dedicated example that illustrates the testing mode ``05_testing_mode.py``

Command line interface

- fixed the command line interface. ``ExperimentCLI`` should now be working with the new pycomex functional API
- Switched to using python ``rich`` package for the CLI printing
- Changed the styling of the "list" and "info" commands to rich formatting

Other changes

- During construction an ``Experiment`` instance will now attempt to automatically parse the parameter description strings from the
module's comments and the parameter typing information from the type hints annotations dict. This information will then be stored
in ``Experiment.metadata`` dictionary.
- Added some more docstrings
- Updated the ``README.rst``
- Added the ``DOCUMENTATION.rst`` and started to compile some additional documentation that is not immediately
relevant to the README

------------------

- I had to change the name of the experiment file copy that is placed in the artifacts folder from "code.py"
to "experiment_code.py" because there was a very weird naming collision with tensorflow internals
- Also adjusted the anylsis file template accordingly.

------------------

- In the functional interface, added the crucial feature of default hook implementations
- Fixed an important bug to make analysis.py files work with sub experiments