feat(nn): [#230] Add MLPClassifier and MLPRegressor models

[vaishnavidesai09]

Summary

Adds MLPClassifier and MLPRegressor under fenn.nn.models.mlp, giving
users a scikit-learn-style API to train basic MLPs without writing Torch
code. Both estimators build a torch.nn.Sequential internally and delegate
all training to the existing ClassificationTrainer / RegressionTrainer —
no training-loop code added.

API

• MLPClassifier(hidden_layer_sizes, activation, solver, learning_rate_init, batch_size, max_iter, early_stopping, n_iter_no_change, validation_fraction, device) — .fit(X, y), .predict(X),
  .predict_proba(X), .score(X, y). Handles binary and multi-class;
  labels don't need pre-encoding.
• MLPRegressor — same constructor, .fit, .predict, .score (R²).
• Both exported from fenn.nn.models and fenn.nn.

Known issue found while testing

ClassificationTrainer.fit()'s validation branch doesn't reshape labels to
(-1, 1) for binary classification the way the training branch does, so
BCEWithLogitsLoss breaks as soon as a val_loader is passed with
num_classes == 2. I didn't fix it here since this module shouldn't contain
training-loop code — flagged as a test marked xfail in
test_early_stopping_runs (TestMLPClassifierBinary) with a full explanation.
Happy to open a separate issue/PR for it if that's preferred.

Testing

20 new unit tests in tests/unit/nn/test_mlp.py, covering binary/multiclass
classification, regression, string labels, predict_proba, error handling,
and early stopping. ruff check, ruff format --check, and typos all pass.

Fixes #230

[vaishnavidesai09]

Hi @blkdmr pls review this pr let me know if changes required

[ApusBerliozi]

Hey @vaishnavidesai09 !

Thx 4 your effort! Kindly check why ty is failing in pre-commit

[vaishnavidesai09]

@ApusBerliozi review it pls

[ApusBerliozi]

Looks fine by me

@blkdmr, gotta wait for your response, because I can't really check by now whether ML side was done correctly

[ApusBerliozi]

@vaishnavidesai09, meanwhile, you can correct one small issue with ruff that shows in pre-commit

[vaishnavidesai09]

@blkdmr pls review this

[ApusBerliozi]

@blkdmr pls review this

Alessio is currently finishing his mater's degree, so I guess he will respond to you at monday or smth