[montionugera]

Closes #4

Summary

• Add FeatureScaler subclassing BasePreprocessor in src/preprocessing_toolkit/scaler.py supporting method="standard" ((x - mean) / std, ddof=0) and method="minmax" ((x - min) / (max - min)).
• Scale numeric columns using training statistics stored in center_ and scale_, pass non-numeric columns through unchanged, preserve missing values, and set scale to 1.0 for constant columns.
• Export FeatureScaler in src/preprocessing_toolkit/__init__.py.
• Add unit and scikit-learn pipeline / GridSearchCV tests in tests/test_scaler.py.

Verification

• ruff check && ruff format --check && pytest

[akifkulll]

Thanks @montionugera, and sorry for the wait. CI didn't run because workflows from first-time contributors need maintainer approval; it's approved now. Locally all 56 tests pass and ruff is clean. The implementation is clean and follows the base contract.
One non-blocking suggestion: a near-constant float column like [0.1+0.2, 0.3, 0.3] gets scale_ ≈ 4.5e-17 instead of 1.0, so floating-point noise is blown up. Treating scales that are tiny relative to the column's magnitude as zero (scikit-learn does something similar) would cover this. Happy to merge once CI is green.

[akifkulll]

thanks!