SFS finalize_fit() support for numpy >= 2.0

[d-kleine]

Description

Update numpy negative infinity constant for version compatibility for the sequential feature selector (SFS)

• Add version check to handle different numpy versions
• Use np.NINF for numpy <2.0 and -np.inf for numpy >=2.0
• Replace direct np.NINF usage with version-aware variable
• Maintain backward compatibility while supporting newer numpy versions

This change ensures the code works correctly across different numpy versions
while maintaining the same functionality.

Related issues or pull requests

fixes #1100 related issue
#1104 similar but incomplete PR

Pull Request Checklist

• Added a note about the modification or contribution to the ./docs/sources/CHANGELOG.md file (if applicable)
• Added appropriate unit test functions in the ./mlxtend/*/tests directories (if applicable)
• Modify documentation in the corresponding Jupyter Notebook under mlxtend/docs/sources/ (if applicable)
• Ran PYTHONPATH='.' pytest ./mlxtend -sv and make sure that all unit tests pass (for small modifications, it might be sufficient to only run the specific test file, e.g., PYTHONPATH='.' pytest ./mlxtend/classifier/tests/test_stacking_cv_classifier.py -sv)
• Checked for style issues by running flake8 ./mlxtend

[d-kleine]

Example code for testing:

import numpy as np
from sklearn.linear_model import LogisticRegression
from mlxtend.feature_selection import SequentialFeatureSelector as SFS

# Generate random values
np.random.seed(42)  # For reproducibility
n_samples = 100     # Number of samples
n_features = 10     # Number of features

# Generate random features (X) and binary target (y)
X_train = np.random.rand(n_samples, n_features)
y_train = np.random.randint(0, 2, size=n_samples)

# Now run Sequential Feature Selection
sfs = SFS(LogisticRegression(),
          k_features=5,          # number of features to select
          forward=True,          # forward selection
          floating=False,        # no floating selection
          scoring='accuracy',
          verbose=5,
          cv=5).fit(X_train, y_train)

[aghents]

Great contribution

[d-kleine]

@rasbt ready for review

[rasbt]

Looks great, thanks for fixing this!