add `preprocessing` + `pipeline` modules

[eonu]

Major changes

• Added sequentia.preprocessing module with sklearn.preprocessing compatibility.
  • Added IndependentFunctionTransformer for applying arbitrary stateless transforms to sequences independently (based on sklearn.preprocessing.FunctionTransformer).
  • Support stateful class-based transforms based on the TransformerMixin interface.
  • Support stateless function-based transforms such as scale and minmax_scale — to be used with IndependentFunctionTransformer.
  • Added mean_filter and median_filter function transforms.
• Added sequentia.pipeline module for sklearn.pipeline compatibility.
  • Added Pipeline for combining transforms and Sequentia estimators into a wrapper that supports sequences as input (based on sklearn.pipeline.Pipeline).
• Upgrade sklearn version specifier from >=0.22 to >=1.0.

Minor changes

• Update README.md example to demonstrate preprocessing and clearer input format.
• Add Scikit-Learn BSD-3 license notices to README.md and various files.