@legitishan In the process a submitting a PR, the changes should solve a single PR such that it would be easier to review and track the history. Could you revert the change for #19913 and open a new PR for these changes.

Regarding solving #19804, it seems that our CI is not happy with this change:

https://dev.azure.com/scikit-learn/scikit-learn/_build/results?buildId=28145&view=logs&j=78a0bf4f-79e5-5387-94ec-13e67d216d6e&t=f1857171-4a53-55c7-3ab5-90acfe091baa&l=380

I am not sure yet if this is our test that is not happy with this our that something else is going on. I will quickly check.

OK so I see that the issue was in MLPRegressor while you modified MLPClassifier. The CI failure is justified.
I also so that a PR has been opened not so long time ago: #19805

I will close this PR in favor of this previous PR that I will review and potentially merge.
Regarding the issue with LOF, you already submitted a PR that solved this issue: #19917

While this is not available yet in the stable version, we can see the change in the dev version:

https://scikit-learn.org/dev/modules/outlier_detection.html

If you wish to solve a new issue, you can git checkout main, git pull (to get the last changes) and git checkout -b new_branch and start solving the new issue. Hope that it helps.

Thanks!