Fixing classifier-related bugs

[drewoldag]

Need to use self.kwargs.pop(foo, bar) in the classifier subclasses so that the key/values are removed from the kwargs dict, and not passed to the classifier instance twice.

Also need to use the correct method name, load_classifier instead of load when using a ResspectClassifier subclass in database.py.

[github-actions]

Before [56fad73]	After [d5da3ad]	Ratio	Benchmark (Parameter)
155±2ms	159±3ms	1.03	benchmarks.time_learn_loop('KNN', 'UncSampling')
195M	197M	1.01	benchmarks.peakmem_learn_loop('KNN')
186M	188M	1.01	benchmarks.peakmem_learn_loop('RandomForest')
151±0.8ms	153±2ms	1.01	benchmarks.time_learn_loop('KNN', 'RandomSampling')
2.59±0.01s	2.62±0.01s	1.01	benchmarks.time_learn_loop('RandomForest', 'RandomSampling')
135±2ms	135±1ms	1	benchmarks.time_feature_creation
2.59±0s	2.56±0.01s	0.99	benchmarks.time_learn_loop('RandomForest', 'UncSampling')

Click here to view all benchmarks.