Add --verify-data CLI flag, Tensor.data property, parallel DataExecutor

- CLI: --verify-data flag enables Phase 2 data verification (ADR-0020)
- Tensor.data: returns actual numpy values (verify-data) or zeros placeholder
- Tensor.__repr__: shows value summary or data=N/A (placeholder)
- DataExecutor: ThreadPoolExecutor for same-timestamp parallel op execution
- BenchResult.engine: exposes op_log/memory_store for Phase 2 access

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

src/kernbench/runtime_api/bench_runner.py

@@ -62,6 +62,7 @@ def run_bench(
             correlation_id=correlation_id,
             trace=None,
             traces=collected_traces,
+            engine=engine,
         )
 
     if completion_policy == CompletionPolicy.LAST_SUBMITTED:
@@ -69,7 +70,7 @@ def run_bench(
         completion, trace = engine.get_completion(last)
         return BenchResult(
             completion=completion, correlation_id=correlation_id,
-            trace=trace, traces=collected_traces,
+            trace=trace, traces=collected_traces, engine=engine,
         )
 
     if completion_policy == CompletionPolicy.ALL_OK_FAIL_FAST:
@@ -80,11 +81,11 @@ def run_bench(
             if not c.ok:
                 return BenchResult(
                     completion=c, correlation_id=correlation_id,
-                    trace=last_trace, traces=collected_traces,
+                    trace=last_trace, traces=collected_traces, engine=engine,
                 )
         return BenchResult(
             completion=Completion(ok=True), correlation_id=correlation_id,
-            trace=last_trace, traces=collected_traces,
+            trace=last_trace, traces=collected_traces, engine=engine,
         )
 
     # LAST_COMPLETED placeholder (needs engine support for timing). Fall back.
@@ -92,5 +93,5 @@ def run_bench(
     completion, trace = engine.get_completion(last)
     return BenchResult(
         completion=completion, correlation_id=correlation_id,
-        trace=trace, traces=collected_traces,
+        trace=trace, traces=collected_traces, engine=engine,
     )