Fix Learner2D state rollback in BalancingLearner ask(tell_pending=False)

[Copilot]

Description

This addresses the failing test (ubuntu-latest, 3.13) job caused by test_balancing_learner[Learner2D-ring_of_fire-learner_kwargs2]. The failure came from state leaking across restore(...) boundaries, eventually leaving Learner2D with too few points during stack fill.

• Root cause

  • Learner2D.__getstate__ returned live mutable objects (_stack, data), so temporary mutations made during BalancingLearner.ask(..., tell_pending=False) were not fully isolated.

• Code change

  • Learner2D.__getstate__ now snapshots _stack and data via deep copies to guarantee rollback correctness when used with restore(...).

• Regression coverage

  • Added a focused test asserting that Learner2D stack/data are unchanged after BalancingLearner.ask(1, tell_pending=False).

def __getstate__(self):
    return (
        cloudpickle.dumps(self.function),
        self.bounds,
        self.loss_per_triangle,
        deepcopy(self._stack),
        deepcopy(self._get_data()),
    )

Checklist

• Fixed style issues using pre-commit run --all (first install using pip install pre-commit)
• pytest passed

Type of change

Check relevant option(s).

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• (Code) style fix or documentation update
• This change requires a documentation update