Isolation Forest

Create a library implementation for the Isolation Forest outlier detection algorithm. I need it to support both the standard Isolation Forest and an Extended Isolation Forest variant that uses random hyperplane splits. The core functionality should allow for distributed training and scoring using Scala and Spark data structures, ideally integrating with Spark's ML library for features like Pipelines. Additionally, please include support for saving and loading trained models, and the ability to export a trained model to ONNX format for cross-platform inference, with a Python example for using the ONNX model.