v0.2.1

This release focuses on reproducibility, documentation upgrades, and better benchmarking utilities, alongside a few behavioral
tweaks and performance improvements in core components.

Highlights:

• Benchmarks & scripts: added/expanded benchmarking scripts and utilities, plus run helpers (.bat/.sh) for scripted runs.
• Docs overhaul: significantly expanded usage docs, new/updated benchmarks page, and refreshed examples/assets (including Titanic
  and PEMS-Bay outputs).
• Imputation/estimators: defaults and preprocessing adjusted (e.g., ridge default intercept behavior, pre‑normalization
  improvements), plus a classifier swap to DecisionTree.
• Performance: reduced repeated pd.concat usage and added timing tests.
• Reproducibility: new/updated reproducibility workflows and assets.