What this PR does / why we need it:

Native MLflow integration for Feast that provides automatic feature lineage tracking alongside ML experiments. When enabled via feature_store.yaml, every feature retrieval is logged to the active MLflow run.

Key Capabilities

1. Auto-logging: Feature retrieval metadata is tagged on the active MLflow run (feast.feature_refs, feast.feature_views, feast.feature_service, feast.entity_count, etc.)
2. Entity DataFrame archival: Optionally saves the training entity DataFrame as an MLflow artifact (entity_df.parquet) for full reproducibility
3. Model-to-feature-service resolution: store.mlflow.resolve_features() maps any MLflow model URI back to its Feast feature service, enabling serving pipelines to auto-discover which features a model needs
4. Entity DataFrame reconstruction: store.mlflow.get_training_entity_df() rebuilds the exact entity DataFrame from a past run's artifacts, enabling training reproducibility
5. Model lifecycle lineage: log_model auto-attaches feast_features.json, register_model propagates feast.feature_service to model versions, load_model links prediction runs to training runs
6. Operations audit trail: feast apply and feast materialize logged to a dedicated {project}-feast-ops experiment
7. Dataset tracking: log_training_dataset() logs training DataFrames as MLflow dataset inputs
8. Two access patterns: store.mlflow (explicit, multi-store safe) and feast.mlflow (drop-in replacement for import mlflow)
9. Configuration: Controlled entirely via feature_store.yaml under a new mlflow: block

Demo

Full end-to-end demo notebook with all capabilities: integration-example

Which issue(s) this PR fixes:

Checks

• I've made sure the tests are passing.
• My commits are signed off (git commit -s)
• My PR title follows conventional commits format

Testing Strategy

• Unit tests
• Integration tests
• Manual tests
• Testing is not required for this change

Misc