API: preprocessing.scaling

skyulf.preprocessing.scaling

Scaling nodes (Standard / MinMax / Robust / MaxAbs).

Each Applier/Calculator dispatches on engine via apply_dual_engine / fit_dual_engine (see dispatcher.py) so per-engine logic lives in small, individually testable _apply_polars / _apply_pandas / _fit_polars / _fit_pandas static helpers. This keeps the public apply / fit methods at CCN 1 (Codacy lizard_ccn-medium).