incorporate mlflow into the ml training evaluation

[bourdeet]

1. look at a typical ml training pipeline
2. identify dashboards / plots of relevance form anomaly detection and output prediction
3. set up a container server up alongside postgres etc.