Add an evaluator for Bias and Fairness metrics

[rogancarr]

We currently have model evaluators that produce metrics on the predicted label. For practical use of machine learning, it is necessary to have a sense for any biases the model may propagate and any fairness issues the mode has. In this way, it would be great to have a standard evaluators for bias and fairness metrics.

Related to #511

[tauheedul]

@rogancarr If you need ideas on how this could be tackled in ML.NET, you may find it useful to research Pymetrics Open Source Audit AI

It is used for Bias Testing of Generalized Machine Learning Applications
Source: https://github.com/pymetrics/audit-ai

[codemzs]

thanks but not on the roadmap for the near future.