Add vLLM v0.18.1 source tree with KV transfer abort fix

third_party/vllm/ now tracked in git for direct patch management.
Based on vLLM v0.18.1 release with one patch applied:

  vllm/v1/core/sched/scheduler.py:
    Replace fatal assert with graceful skip when KV transfer callback
    arrives for an already-aborted request during PD disaggregated serving.

Future vLLM modifications should be made directly in third_party/vllm/
and committed normally. The patches/ directory is kept as documentation
of what changed from upstream.

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

third_party/vllm/tests/distributed/eplb_utils.py

@@ -0,0 +1,54 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+import os
+import random
+
+import torch
+import torch.multiprocessing as mp
+
+from vllm.config import VllmConfig, set_current_vllm_config
+from vllm.distributed.parallel_state import (
+    init_distributed_environment,
+)
+from vllm.utils.system_utils import update_environment_variables
+
+mp.set_start_method("spawn", force=True)
+
+
+def distributed_run(fn, world_size, *args):
+    number_of_processes = world_size
+    processes: list[mp.Process] = []
+    for i in range(number_of_processes):
+        env: dict[str, str] = {}
+        env["RANK"] = str(i)
+        env["LOCAL_RANK"] = str(i)
+        env["WORLD_SIZE"] = str(number_of_processes)
+        env["LOCAL_WORLD_SIZE"] = str(number_of_processes)
+        env["MASTER_ADDR"] = "localhost"
+        env["MASTER_PORT"] = "12345"
+        p = mp.Process(target=fn, args=(env, world_size, *args))
+        processes.append(p)
+        p.start()
+
+    for p in processes:
+        p.join()
+
+    for p in processes:
+        assert p.exitcode == 0
+
+
+def set_env_vars_and_device(env: dict[str, str]) -> None:
+    update_environment_variables(env)
+    local_rank = os.environ["LOCAL_RANK"]
+    device = torch.device(f"cuda:{local_rank}")
+    torch.accelerator.set_device_index(device)
+
+    # Create a minimal vllm config for init_distributed_environment
+    vllm_config = VllmConfig()
+    with set_current_vllm_config(vllm_config):
+        init_distributed_environment()
+
+    # Ensure each worker process has the same random seed
+    random.seed(42)
+    torch.manual_seed(42)