Enable QSVC and VQC baselines in quantum classification example

[Copilot]

examples/quantum/classification.py had the QSVC (QSVM) and VQC (QNN) Qiskit baselines commented out. This re-enables them so they run alongside the classical and quantum HD models under the existing K-fold harness.

Note: the request mentioned "VQE"; the file only contains a VQC (Variational Quantum Classifier) — no VQE code exists. Enabled VQC on that assumption.

Changes (all in examples/quantum/classification.py)

• Imports: uncommented COBYLA, ZFeatureMap, ZZFeatureMap, RealAmplitudes, VQC, QSVC, ComputeUncompute, FidelityQuantumKernel, StatevectorSampler.
• run_fold: restored the QSVC block (ZZFeatureMap + FidelityQuantumKernel via StatevectorSampler/ComputeUncompute, fit/predict, timing, report, confusion matrix, decision-function ROC data) and the VQC block (ZFeatureMap + RealAmplitudes ansatz + COBYLA(maxiter=100), fit/predict, neural_network.forward for class-1 scores).
• Checkpointing: re-added vqc_* / qsvc_* entries to the per-fold fold_data JSON payload.
• Aggregation: re-added the vqc_reports/matrices/times/roc_data and qsvc_* lists plus their per-fold append/extend in the deterministic-order loop.
• Reporting: re-added the print_cv_summary calls, raw ROC data prints, and roc_curve-based (FPR, TPR) plot-point sections for both models.