fix(binning): make get_feature_names_out follow the sklearn contract

[ChrisW09]

Summary

CustomBinTransformer.get_feature_names_out() raises when called with no arguments and
returns a plain list rather than an ndarray. Both break the scikit-learn contract, and the
first makes Pipeline.get_feature_names_out() fail for any pipeline containing the
transformer.

The package already fixed exactly this for three sibling transformers — see the docstring of
tests/test_feature_names_out.py ("get_feature_names_out(None) defaults to generated
x0, x1, ... names instead of raising, for NoTransformer, ToFloatTransformer and
ContinuousOrdinalTransformer") — but CustomBinTransformer was left out.

Reproduction

import numpy as np
from sklearn.pipeline import Pipeline
from pretab.transformers import CustomBinTransformer, NoTransformer

X = np.linspace(0, 1, 20).reshape(-1, 1)
t = CustomBinTransformer(output_dim=4).fit(X)

t.get_feature_names_out()

InvalidParamError: input_features must be specified

print(type(t.get_feature_names_out(["f"])).__name__)                     # list
print(type(NoTransformer().fit(X).get_feature_names_out()).__name__)     # ndarray

Pipeline([("bin", CustomBinTransformer(output_dim=4))]).fit(X).get_feature_names_out()

InvalidParamError: input_features must be specified

Expected

Consistent with every other transformer in the package and with scikit-learn:
get_feature_names_out() works with no arguments, generating x0, x1, ..., and always
returns an ndarray of strings.

Actual

Raises without arguments; returns a list with them.

Root cause

pretab/transformers/binning/binning.py:142-157:

def get_feature_names_out(self, input_features=None):
    if input_features is None:
        raise InvalidParamError("input_features must be specified")
    return input_features

It also does not guard on being fitted, unlike the siblings which call check_is_fitted.

Suggested fix

Mirror NoTransformer.get_feature_names_out:

def get_feature_names_out(self, input_features=None):
    check_is_fitted(self, "n_features_in_")
    if input_features is None:
        input_features = [f"x{i}" for i in range(self.n_features_in_)]
    return np.asarray(input_features, dtype=object)

Note this changes two existing tests that pin the current behaviour:
test_custom_bin_transformer_feature_names_out_raises (asserts the no-argument call raises)
and test_custom_bin_transformer_feature_names_out (asserts a list is returned, via
names == ["feature1"]). Both encode the defect rather than a desired guarantee.

Environment

• pretab 0.1.0 (main @ 51c3043)
• Python 3.11.15, numpy 2.4.6, pandas 2.3.3, scikit-learn 1.9.0, scipy 1.17.1
• macOS (darwin 25.5.0)