feat: accept metric names without neg_ prefix in summarize()

[direkkakkar319-ops]

Description

Fixes #2607
When using report.metrics.summarize(), passing error metric names without the neg_ prefix (e.g. "mean_squared_error") would raise a ValueError because sklearn's scorer registry only knows the neg_ .

Root Cause

Sklearn follows a "higher is better" convention for all scorers.
Error metrics like mean_squared_error are therefore only registered as neg_mean_squared_error in sklearn's scorer registry.

Solution

Added a _METRIC_ALIASES dictionary inside _MetricsAccessor that maps user-friendly metric names to their neg_ prefixed sklearn scorer equivalents.

_METRIC_ALIASES: dict[str, str] = {
    "mean_squared_error": "neg_mean_squared_error",
    ...
}

Changes

• skore/src/skore/_sklearn/_estimator/metrics_accessor.py — added _METRIC_ALIASES dict and resolution logic in summarize()
• In skore/src/skore/_sklearn/_estimator/metrics_accessor.py
  made this code

to

• skore/tests/unit/reports/estimator/metrics/test_metric_aliases.py — added test cases for the fix
• sphinx/changelog.rst — updated changelog

Testing

# Before this fix — raised ValueError
report.metrics.summarize(metric='mean_squared_error').frame()

# After this fix-working correct
report.metrics.summarize(metric='mean_squared_error').frame()
report.metrics.summarize(metric='neg_mean_squared_error').frame()  # still works too