Kernel-launch sync (ADR-0009 D5) and IPCQ drain at inbound (ADR-0023)

- KernelLaunchMsg gains target_start_ns: IO_CPU stamps a global barrier
  (max path latency across every target PE), M_CPU passes it through,
  PE_CPU yields until it before recording pe_exec_start. Every PE in a
  launch begins kernel execution at the same env.now regardless of its
  dispatch path length — eliminates per-PE dispatch-offset artifact in
  cross-PE and cross-cube latency measurements.

- PE_DMA._handle_ipcq_inbound now pays Transaction.drain_ns at the top,
  matching the terminal-drain behavior of ComponentBase._forward_txn for
  every non-IPCQ Transaction. SRC-side tl.send stays fire-and-forget
  (sender doesn't yield on sub_done); tl.recv now blocks until bytes
  have actually drained into its inbox.

- ComponentContext: new compute_path_latency_ns helper + node_overhead_ns
  field populated by GraphEngine.

- tests/test_kernel_launch_sync.py: asserts all PEs in one launch
  produce identical pe_exec_ns for a no-op kernel (zero spread).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/kernbench/runtime_api/kernel.py

@@ -90,6 +90,11 @@ class KernelLaunchMsg:
     args: tuple[KernelArg, ...]
     target_cubes: tuple[int, ...] | Literal["all"] = "all"
     target_pe: int | tuple[int, ...] | Literal["all"] = "all"
+    # ADR-0009 D5: synchronized kernel start. When set, each PE_CPU yields
+    # until env.now >= target_start_ns before beginning kernel execution,
+    # so every PE in a launch starts at the same simulated time regardless
+    # of its M_CPU dispatch path length. Stamped by M_CPU fan-out.
+    target_start_ns: float | None = None
     msg_type: Literal["kernel_launch"] = "kernel_launch"