DOC Example MLPRegressor as an autoencoder

[lfourniercm]

Reference Issues/PRs

Fixes #25346
Fixes #25216

What does this implement/fix? Explain your changes.

This fix adds an example for MLPRegressor as an autoencoder

Any other comments?

[github-actions]

✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

_{Generated for commit: 002c6fc. Link to the linter CI: here}

[glemaitre]

@ogrisel I just push my improved example. However, I am wondering if we should actually add this example since in a real use-case we would advocate to use a dedicated neural network library (e.g. keras) to tackle such problem.

I would second your opinion either way on this one.

[jonnor]

Denoising an image is maybe not a great example. Especially since very quickly, one probably would want to use convolutions etc., which are outside sklearn scope. But using a reconstructing autoencoder as an anomaly detection method is quite established, and an improvement over OneClassSVM/IsolationForest/LocalOutlierFactor etc in many cases. Maybe a basic example of that would be appropriate?
Similar to ex https://scikit-learn.org/stable/auto_examples/neighbors/plot_lof_novelty_detection.html#sphx-glr-auto-examples-neighbors-plot-lof-novelty-detection-py ?