Machine Learning Feature Store with Temporal Consistency

Create a Python feature store implementation that guarantees temporal consistency for machine learning model training and inference. Design a system that supports point-in-time correct feature retrieval, handling time-travel queries, and managing feature versioning. Implement efficient storage strategies using columnar formats like Parquet, support for both batch and streaming feature generation, and automatic feature drift detection. Include a pluggable feature transformation pipeline with lineage tracking and metadata management.