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_framework.py

@@ -0,0 +1,62 @@
+"""Tests for the torch.distributed-compat facade (ADR-0023 D11).
+
+These tests verify the public API surface of ``DistributedContext`` +
+``AhbmCCLBackend``. End-to-end correctness of the allreduce itself is
+covered by tests/test_ccl_allreduce_matrix.py.
+"""
+from __future__ import annotations
+
+from kernbench.runtime_api.distributed import AhbmCCLBackend, DistributedContext
+
+
+def test_init_process_group_requires_ctx_ref():
+    """Using DistributedContext without RuntimeContext binding should fail."""
+    dist = DistributedContext()
+    # Not bound to a RuntimeContext → init should raise.
+    try:
+        dist.init_process_group(backend="ahbm")
+        assert False, "expected RuntimeError"
+    except RuntimeError:
+        pass
+
+
+def test_init_process_group_rejects_unknown_backend():
+    """Unknown backend raises ValueError (matches pytorch behavior)."""
+    dist = DistributedContext()
+    dist._ctx_ref = object()  # dummy; won't be reached before the check
+    try:
+        dist.init_process_group(backend="nccl")
+        assert False, "expected ValueError"
+    except ValueError:
+        pass
+
+
+def test_distributed_pytorch_compat_surface():
+    """DistributedContext only exposes real torch.distributed API names."""
+    # Every public attribute should either be a real pytorch name or private.
+    allowed = {
+        "init_process_group",
+        "is_initialized",
+        "get_world_size",
+        "get_rank",
+        "get_backend",
+        "all_reduce",
+        "barrier",
+    }
+    dc = DistributedContext()
+    for attr in dir(dc):
+        if attr.startswith("_"):
+            continue
+        assert attr in allowed, (
+            f"DistributedContext exposes non-pytorch API: {attr!r}"
+        )
+
+
+def test_backend_class_surface():
+    """AhbmCCLBackend exposes only all_reduce + barrier + world_size."""
+    # Ensure we don't accidentally leak internal method names.
+    public = {m for m in dir(AhbmCCLBackend) if not m.startswith("_")}
+    # Class must at minimum expose these.
+    assert "all_reduce" in public
+    assert "barrier" in public
+    assert "world_size" in public