feat(nn): [#220] add basic SLP classifier and regressor

[NoumanParvez12]

Implements SLPClassifier and SLPRegressor in fenn/nn/models/slp.py, following the structure of MLPClassifier/MLPRegressor in mlp.py, as discussed with @blkdmr on the issue.

Mirrors sklearn.linear_model.Perceptron's API for classification; regression (SLPRegressor) follows the same pattern since sklearn has no direct equivalent for regression.
Delegates all training to the existing ClassificationTrainer/RegressionTrainer — no training logic added, per issue requirements.
Naming follows maintainer guidance (SLP prefix, not the full "SingleLayerPerceptron" name)
_build_slp builds a single nn.Linear layer with no hidden layers/activation, unlike _build_mlp
Added tests/unit/nn/test_slp.py, mirroring test_mlp.py's structure — all tests pass, full existing suite (936 tests) still passes.

Closes #220

[NoumanParvez12]

@blkdmr This is ready for review. A couple of notes:

Since sklearn.linear_model.Perceptron uses penalty/alpha/eta0-style params but our Trainer classes are built around a PyTorch optimizer, I kept solver/learning_rate_init (matching MLPClassifier's pattern) rather than mirroring sklearn's exact param names. Happy to adjust if you'd rather match sklearn more closely.
Let me know if you'd prefer a different file/class naming split, or if anything else should change before merge.