OD-Issues

[mauicv]

This issue tracks the issues relating to the outlier detector implementation:

Quick stuff:

• Refactor device types to _types for outlier detectors #779
• Remove duplicated utils._types logic #763
• Threshold datatype #809
• Set meta data on all OD #811
• Docstrings/Naming for _to_numpy method in SKLearn backend #812
• Return data from od fit methods #813
• kNN detector defaults #760
• gmm and svm set multiple defaults #819
• Add arguments and return types protocols #755

Long Stuff:

• Normalizer and Aggregator Docstrings #805
• Args for ODs methods with multiple backends #787
• Use DataClasses to type detector return values #762
• Review data_type in meta #807
• Track Skipped OD tests #810
• Add serialisation new outlier detectors #831

Final Stuff:

• Error checks in deployment for kNN detector #761
• Replace anomalous with outlying throughout #808
• Check consistent order of arguments in Outlier Detectors #818