fix: SklearnModelInfo.target_values returns both classes for binary classifiers

[agu2347]

Problem

Registering a binary scikit-learn classifier via register_model() raises:

ValueError: Please provide all possible values for the target variable, including a no-event value.

Root cause

SklearnModelInfo.target_values (in src/sasctl/utils/model_info.py) returns only the single "positive" class for binary classifiers:

@property
def target_values(self):
    if self.is_binary_classifier:
        return [self.model.classes_[-1]]
    if self.is_classifier:
        return list(self.model.classes_)

_register_open_source_model (in tasks.py) passes this straight through to JSONFiles.write_model_properties_json(target_values=info.target_values), which explicitly requires either 0, 2, or >2 target values — and raises a ValueError for exactly 1:

elif isinstance(target_values, list) and len(target_values) == 2:
    ...
else:
    raise ValueError(
        "Please provide all possible values for the target variable, including"
        " a no-event value."
    )

So any binary classifier hits this branch and fails. info.target_values is also passed to ImportModel().import_model(target_values=info.target_values, ...), whose docstring says "the order of the target values should reflect the order of the related probabilities in the model" — i.e. it expects both classes too.

Closes #200

Fix

Return both classes for binary classifiers, in classes_ order ([non-event, event]), matching the ordering already used for output_column_names (e.g. ["P_benign", "P_malignant"]) and predict_proba columns.

Testing

• Updated the existing test_sklearn_binary_classifier assertion, which previously pinned the single-value (buggy) behavior, to expect both classes.
• Added test_sklearn_binary_classifier_target_values_registration, which reproduces the exact reported crash by calling write_model_properties_json with info.target_values for a fitted binary classifier and asserting it no longer raises.
• Confirmed the new/updated tests fail with the reported error when the fix is reverted (stashed model_info.py only, tests unchanged): AssertionError: assert ['malignant'] == ['benign', 'malignant'] / assert 1 == 2.
• All 18 tests in tests/unit/test_model_info_sklearn.py pass with the fix applied.
• black --check passes on both changed files.