Variational option discovery algorithms capability update

AI Impact Summary

Introducing variational option discovery algorithms signals a new capability in the RL toolkit, enabling automatic discovery of reusable sub-policies via variational methods. This could shift exploration strategies, improve sample efficiency, and affect benchmarking results for hierarchical RL workloads. Teams should review API docs for new methods, adjust experiment configs (e.g., priors, KL penalties), and run controlled comparisons to quantify performance changes before adopting in production pipelines.