Add PE-level IPCQ collective infra + unified ccl_allreduce bench (ADR-0023)

Major changes:

PE-level IPCQ infrastructure:
- New PE_IPCQ component: ring-buffer control plane with 4-direction
  neighbor mapping, head/tail pointers, backpressure (poll/sleep).
- PE_DMA extended with vc_comm channel for IPCQ outbound/inbound DMA,
  including in-flight data snapshot (D9) and op_log recording at
  outbound time for Phase 2 replay correctness.
- IpcqDmaToken piggyback model: data + metadata travel together,
  atomic visibility at receiver (invariant I6).
- Credit return fast path: bottleneck-BW latency, no fabric vc_comm.

Phase 2 data execution (ADR-0020 integration):
- op_log extended: DmaWriteCmd now captures src_space/src_addr for
  Phase 2 dma_write copy; ipcq_copy ops recorded at outbound time.
- DataExecutor replays dma_write + ipcq_copy in t_start order.
- Engine._flush_data_phase: incremental cursor-based replay after
  each engine.wait() so host reads see post-Phase-2 data.
- KernelRunner Phase 1 writes disabled when op_log is active to
  prevent stale data from corrupting the MemoryStore snapshot.

TLContext / kernel API:
- tl.send(dir, src=TensorHandle), tl.recv(dir, shape, dtype),
  tl.recv_async, tl.wait(RecvFuture), copy_to_dst mode.
- TensorHandle operator overloading (add/sub/mul/div) via thread-local
  active TLContext → MathCmd dispatch through PE_MATH.
- PE-local scratch allocator for math output handles.
- tl.load returns space="hbm" handles for correct Phase 2 addressing.
- Additional math functions: maximum, minimum, fma, clamp, softmax, cdiv.

Unified ccl_allreduce bench (PyTorch-compat host code):
- Single benches/ccl_allreduce.py with run() + worker(rank, ws, torch)
  split matching real PyTorch DDP worker pattern.
- torch.distributed facade: init_process_group, get_world_size,
  get_rank, get_backend, all_reduce, barrier — only real PyTorch names.
- AhbmCCLBackend: eager install_ipcq at init, all_reduce dispatches
  kernel via tensor shard metadata (n_elem from shards[0].nbytes).
- world_size derived from topology spec (sips × cubes × pes_per_cube)
  with optional algorithm-level override in ccl.yaml.

Tensor API (PyTorch-compat surface):
- Tensor.numpy(): gather-aware (all shards via VA-based addressing).
- Tensor.copy_(source): scatter from host tensor into sharded target.
- RuntimeContext.from_numpy(arr): host-side staging tensor.
- Tensor.data property fixed to use numpy() (was shards[0]-only).

Algorithm modules moved to src/kernbench/ccl/algorithms/:
- ring_allreduce, mesh_allreduce, tree_allreduce, hello_send.
- Each module exports kernel_args(world_size, n_elem) helper.
- ccl.yaml module paths updated to kernbench.ccl.algorithms.*.

Dead code removed:
- 7 per-variant bench files (ccl_allreduce_{tcm,hbm,sram}, etc.).
- _run_ccl_bench greenlet-per-SIP scheduler.
- benches.loader.is_ccl_bench + run_rank detection.
- benches/ccl/ directory.

Tests:
- New test_ccl_allreduce_matrix.py: 7 parametrized cases
  (ring×3 buffers, ring 8/16, mesh 4, tree 7).
- New test_runtime_api_tensor.py: copy_/numpy/from_numpy unit tests.
- Existing tests updated for new import paths + world_size_override.

Docs:
- Korean ccl-author-guide.md and ADR-0023 paths updated.
- New English versions: ccl-author-guide.en.md, ADR-0023.en.md.

502 tests pass.

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

tests/test_ccl_diagnostics.py

@@ -0,0 +1,70 @@
+"""Tests for CCL diagnostics: trace + pointer dump (ADR-0023 D14)."""
+from __future__ import annotations
+
+import os
+
+from kernbench.ccl import diagnostics
+
+
+# ── trace toggle ─────────────────────────────────────────────────────
+
+
+def test_trace_disabled_by_default(monkeypatch):
+    monkeypatch.delenv("KERNBENCH_CCL_TRACE", raising=False)
+    diagnostics.reload_trace_setting()
+    assert diagnostics.trace_enabled() is False
+
+
+def test_trace_enabled_via_env(monkeypatch):
+    monkeypatch.setenv("KERNBENCH_CCL_TRACE", "1")
+    diagnostics.reload_trace_setting()
+    assert diagnostics.trace_enabled() is True
+
+
+def test_trace_record_send(monkeypatch, capsys):
+    monkeypatch.setenv("KERNBENCH_CCL_TRACE", "1")
+    diagnostics.reload_trace_setting()
+    diagnostics.log_send(t_ns=100.0, sender="sip0.cube0.pe0",
+                         direction="E", nbytes=64, sender_seq=0)
+    out = capsys.readouterr().out
+    assert "send" in out
+    assert "sip0.cube0.pe0" in out
+    assert "dir=E" in out
+    monkeypatch.delenv("KERNBENCH_CCL_TRACE")
+    diagnostics.reload_trace_setting()
+
+
+def test_trace_record_recv(monkeypatch, capsys):
+    monkeypatch.setenv("KERNBENCH_CCL_TRACE", "1")
+    diagnostics.reload_trace_setting()
+    diagnostics.log_recv(t_ns=200.0, receiver="sip0.cube0.pe1",
+                         direction="W", nbytes=64)
+    out = capsys.readouterr().out
+    assert "recv" in out
+    assert "sip0.cube0.pe1" in out
+    monkeypatch.delenv("KERNBENCH_CCL_TRACE")
+    diagnostics.reload_trace_setting()
+
+
+# ── pointer dump ────────────────────────────────────────────────────
+
+
+def test_pointer_dump_format():
+    from kernbench.sim_engine.engine import GraphEngine
+    from kernbench.topology.builder import resolve_topology
+    from kernbench.ccl.install import (
+        install_ipcq, load_ccl_config, resolve_algorithm_config,
+    )
+
+    topo = resolve_topology("topology.yaml").topology_obj
+    engine = GraphEngine(topo, enable_data=True)
+    cfg = resolve_algorithm_config(load_ccl_config(), name="ring_allreduce_tcm")
+    install_ipcq(engine, topo.spec, cfg)
+
+    dump = diagnostics.pointer_dump(engine)
+    # 8 ranks × 2 directions = 16 lines (plus 8 PE headers)
+    assert "sip0.cube0.pe0" in dump
+    assert "E:" in dump
+    assert "W:" in dump
+    assert "my_head=" in dump
+    assert "peer_tail_cache=" in dump