ENH: Add adjusted_mutual_info_score metric

[vardhan30016]

Summary

This PR implements the adjusted_mutual_info_score metric in cuML’s cluster metrics module.
It mirrors the scikit-learn API for consistent behavior and prepares for future GPU acceleration.

Related Issue

Fixes: #7294

Implementation Details

• Added new file: python/cuml/metrics/cluster/adjusted_mutual_info_score.py
• Updated __init__.py to expose the new metric
• Added docstring with parameters and return type
• Uses sklearn’s implementation internally for now (compatibility wrapper)

Checklist

• Code builds locally
• Unit tested manually
• Follows cuML’s code style
• Linked to the corresponding issue

Author: Vardhan Naidu (@vardhan30016)

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[vardhan30016]

Rebased this branch with the latest main and force-pushed to update the commit history.
All local tests pass successfully, and the CI should now rerun with the latest base.

The implementation for adjusted_mutual_info_score is stable and aligned with scikit-learn’s behavior.

@bdice @csadorf @divyegala — please review when convenient. Thanks for your guidance and feedback!

[csadorf]

Have a look at the diff and commits on this PR, something must have gone wrong with the rebase.

[csadorf]

This PR has significant file organization issues that need cleanup. The same code appears multiple times across different files - including files with invalid extensions like .pyes and .p, files in the repository root that should be in proper package locations, and test files in both correct and incorrect directories. There are also empty placeholder files and an __init__.py marked as new but containing comments like # existing imports... suggesting it should be modifying an existing file instead.

The PR also includes both a CPU-based sklearn wrapper and a separate GPU implementation without clarity on which is intended or how they integrate. Given the extent of cleanup needed, I'm going to close this PR. Please open a new PR with only the necessary files in their correct locations and a clear implementation approach.