[INFRA] Import NVIDIA/CCCL upstream as optimization reference library

CCCL (CUDA C++ Core Libraries) provides:
- CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk)
- Thrust: high-level parallel algorithms (transform_reduce, sort, scan)
- libcudacxx: CUDA C++ standard library (atomics, barriers, memory)
- cudax: experimental features (memory resources, allocators)
- Tuning policies: per-SM hardware-specific algorithm parameters

Competition optimization vectors mapped to CCCL:
- Output TPS (83% weight): warp_reduce, block_reduce, device_topk
- Input TPS (14% weight): device_scan, block_load, prefetch
- Cache TPS (3% weight): prefix caching strategy patterns
- Memory (0.9 util): pooled/cached/buddy allocators

Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only)
License: Apache-2.0

cccl_upstream/cub/cub/util_macro.cuh

@@ -0,0 +1,67 @@
+// SPDX-FileCopyrightText: Copyright (c) 2011, Duane Merrill. All rights reserved.
+// SPDX-FileCopyrightText: Copyright (c) 2011-2024, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
+
+/******************************************************************************
+ * Common C/C++ macro utilities
+ ******************************************************************************/
+
+#pragma once
+
+#include <cuda/__cccl_config>
+
+#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
+#  pragma GCC system_header
+#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
+#  pragma clang system_header
+#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
+#  pragma system_header
+#endif // no system header
+
+#include <cub/detail/detect_cuda_runtime.cuh> // IWYU pragma: export
+#include <cub/util_namespace.cuh> // IWYU pragma: export
+
+CUB_NAMESPACE_BEGIN
+
+#ifdef _CCCL_DOXYGEN_INVOKED
+#  define CUB_DISABLE_KERNEL_VISIBILITY_WARNING_SUPPRESSION
+#endif
+
+/**
+ * @def CUB_DISABLE_KERNEL_VISIBILITY_WARNING_SUPPRESSION
+ * If defined, the default suppression of kernel visibility attribute warning is disabled.
+ */
+#if !defined(CUB_DISABLE_KERNEL_VISIBILITY_WARNING_SUPPRESSION)
+_CCCL_DIAG_SUPPRESS_GCC("-Wattributes")
+_CCCL_DIAG_SUPPRESS_CLANG("-Wattributes")
+#  if !_CCCL_CUDA_COMPILER(NVHPC)
+_CCCL_DIAG_SUPPRESS_NVHPC(attribute_requires_external_linkage)
+#  endif // !_CCCL_CUDA_COMPILER(NVHPC)
+#endif // !CUB_DISABLE_KERNEL_VISIBILITY_WARNING_SUPPRESSION
+
+#ifndef CUB_DEFINE_KERNEL_GETTER
+#  define CUB_DEFINE_KERNEL_GETTER(name, ...)                                               \
+    _CCCL_HIDE_FROM_ABI CUB_RUNTIME_FUNCTION static constexpr decltype(&__VA_ARGS__) name() \
+    {                                                                                       \
+      return &__VA_ARGS__;                                                                  \
+    }
+#endif
+
+// TODO(bgruber): drop in CCCL 4.0 when we drop the public dispatchers
+#ifndef CUB_DEFINE_SUB_POLICY_GETTER
+#  define CUB_DEFINE_SUB_POLICY_GETTER(name)                            \
+    _CCCL_HOST_DEVICE static constexpr auto name()                      \
+    {                                                                   \
+      return MakePolicyWrapper(typename StaticPolicyT::name##Policy()); \
+    }
+#endif
+
+#if defined(CUB_DEFINE_RUNTIME_POLICIES)
+#  define CUB_DETAIL_STATIC_ISH_ASSERT(expr, msg) _CCCL_ASSERT(expr, msg)
+#  define CUB_DETAIL_CONSTEXPR_ISH
+#else // ^^^ CUB_DEFINE_RUNTIME_POLICIES ^^^ / vvv !CUB_DEFINE_RUNTIME_POLICIES vvv
+#  define CUB_DETAIL_STATIC_ISH_ASSERT(expr, msg) static_assert(expr, msg);
+#  define CUB_DETAIL_CONSTEXPR_ISH                constexpr
+#endif // !(CUB_DEFINE_RUNTIME_POLICIES)
+
+CUB_NAMESPACE_END