Simplify Qiskit ParallelExperiment: remove src/ dependency, fix measurement and noise bugs

  - Replace custom partitioning/mapping (parallel_experiment/src/) with simple
    sequential qubit allocation — reduces ibm_fez execution from 200s to 17s
  - Fix measure_all() bug: preserve original circuit measurements instead of
    measuring all qubits (was adding extra bits, breaking fidelity calculation)
  - Fix simulator noise: wrap qasm_simulator in AerSimulator with noise model
    so ParallelExperiment's internal SamplerV2 picks up QED-C noise settings
  - Fix cregs[0].size → num_qubits in 4 files under parallel_experiment/src/
  - Add traceback printing on parallel experiment failure
  - Old custom partitioning preserved as _run_qiskit_parallel_experiment_with_custom_partitioning

Author: rtvuser1

qedclib/qiskit/execute.py

@@ -1404,6 +1404,11 @@ def process_circuit_results(circuits_info, results, job_id=None, elapsed_time=No
         # Call the benchmark's result handler (computes fidelity etc.)
         # Skip if counts are empty (cancelled/failed job) to avoid division-by-zero in handlers
         if result_handler and counts:
+            # DEBUG: show counts reaching result_handler
+            sample = dict(list(counts.items())[:3])
+            print(f"... [debug pcr] circuit {ci['group']}/{ci['circuit']}: "
+                  f"counts_type={type(counts).__name__}, "
+                  f"sample={sample}")
             try:
                 result_handler(ci["qc"], circuit_result,
                               ci["group"], ci["circuit"], ci["shots"])