Implement pluggable classifiers for RESSPECT

[drewoldag]

This PR represents the work to implement a pluggable system for classifiers.

From the users perspective, the API hasn't changed. To specify a classifier, define a LoopConfiguration instance with the classifier value set to one of the preexisting models: "RandomForest", "KNN", "MLP", "SVM", "NBG".

To make use of a classifier defined outside of Resspect, set the classifier value in the LoopConfiguration instance to be the import specification for the external classifier: "external_package.sub_module.ClassifierClassName".

The externally defined classifier will need to inherit from ResspectClassifier like so:

from resspect.classifiers import ResspectClassifier

class ExternalClassifer(ResspectClassifier):
    def __init__(self, train_data, train_labels, test_data, **kwargs):
        super().__init__(train_features, train_labels, test_features, **kwargs)
        self.clf = <The external classifier definition>

• Includes a base ResspectClassifier class and early RandomForest class
• Small edit to ReFitVotingClassifier default parameters
• Implemented RandomForestClassifier as first subclass of ResspectClassifier
• Basic plugin system working that will generate a dictionary of {<class_name>: } that can be used to create a particular class instance at runtime.
• Plugin system also supports loading external subclasses ResspectClassifier by using the import specification string, module.submodule.class_name.
• Implemented ResspectClassifier subclasses for all other built-in classifiers
• Removed if/else ladder code from database.py that is now redundant.
• Added tests for coverage of the plugin system

[github-actions]

Before [a5c71dd]	After [d5ffa21]	Ratio	Benchmark (Parameter)
165±0.7ms	169±3ms	1.02	benchmarks.time_learn_loop('KNN', 'UncSampling')
134±2ms	136±1ms	1.01	benchmarks.time_feature_creation
163±0.5ms	164±2ms	1.01	benchmarks.time_learn_loop('KNN', 'RandomSampling')
173M	173M	1	benchmarks.peakmem_learn_loop('KNN')
169M	169M	1	benchmarks.peakmem_learn_loop('RandomForest')
2.59±0.03s	2.56±0.01s	0.99	benchmarks.time_learn_loop('RandomForest', 'RandomSampling')
2.63±0.03s	2.60±0.02s	0.99	benchmarks.time_learn_loop('RandomForest', 'UncSampling')

Click here to view all benchmarks.