Fix OPTICS min_samples casting to int to avoid NN TypeError (swev-id: scikit-learn__scikit-learn-14496)

[casey-brooks]

Summary

• normalize OPTICS min_samples and min_cluster_size via ceil/clip and reuse ints throughout the graph build
• surface the normalized values on the estimator, cover fractional inputs in docstrings, and update the changelog referencing issue [BUG] Optics float min_samples NN instantiation #41 and Emerson’s research
• add regression tests for fractional and invalid parameters plus protection in cluster_optics_xi

Testing

• LD_LIBRARY_PATH=/nix/store/qipd93x9gjyiygqk673rd2ssnf8y7jj0-gcc-14.3.0-lib/lib:/nix/store/gh2dd8vimringn726ndall19gbm77prj-openblas-0.3.30/lib:/nix/store/y6zrr7dfg051mz4dpjvaldy4g9cy6wmq-lapack-3/lib:/nix/store/4wdz42ns29ys6fm1xak68bnp51nxhd2s-zlib-1.3.1/lib: /workspace/sklearn-py37/bin/python -m pytest sklearn/cluster/tests/test_optics.py
• /workspace/sklearn-py37/bin/flake8 --ignore=E121,E123,E126,E24,E704,W503,W504,E731,E303 sklearn/cluster/optics_.py sklearn/cluster/tests/test_optics.py

Resolves #41.

[casey-brooks]

Test & Lint

• LD_LIBRARY_PATH=/nix/store/qipd93x9gjyiygqk673rd2ssnf8y7jj0-gcc-14.3.0-lib/lib:/nix/store/gh2dd8vimringn726ndall19gbm77prj-openblas-0.3.30/lib:/nix/store/y6zrr7dfg051mz4dpjvaldy4g9cy6wmq-lapack-3/lib:/nix/store/4wdz42ns29ys6fm1xak68bnp51nxhd2s-zlib-1.3.1/lib: /workspace/sklearn-py37/bin/python -m pytest sklearn/cluster/tests/test_optics.py (49 passed)
• /workspace/sklearn-py37/bin/flake8 --ignore=E121,E123,E126,E24,E704,W503,W504,E731,E303 sklearn/cluster/optics_.py sklearn/cluster/tests/test_optics.py (no issues)

[noa-lucent]

Looks good to me. Fractional parameters are now normalized before building the neighbor graph, downstream helpers reuse the integer counts, and the regression plus validation tests catch the prior TypeError scenario. Docs and what's new entries are in place.