[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/cudax/include/cuda/experimental/__device/logical_device.cuh

@@ -0,0 +1,140 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of CUDA Experimental in CUDA C++ Core Libraries,
+// under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+// SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _CUDAX__DEVICE_LOGICAL_DEVICE_CUH
+#define _CUDAX__DEVICE_LOGICAL_DEVICE_CUH
+
+#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 <cuda/__device/all_devices.h>
+#include <cuda/__device/physical_device.h>
+
+#include <cuda/experimental/__green_context/green_ctx.cuh>
+
+#include <cuda/std/__cccl/prologue.h>
+
+namespace cuda::experimental
+{
+struct __logical_device_access;
+
+//! @brief A non-owning representation of a CUDA device or a green context
+class logical_device
+{
+public:
+  //! @brief Enum to indicate the kind of logical device stored
+  enum class kinds
+  {
+    // Indicates logical device is a full device
+    device,
+    // Indicated logical device is a green context
+    green_context
+  };
+
+  // We might want to make this private depending on how this type ends up looking like long term,
+  // not documenting it for now
+  [[nodiscard]] constexpr CUcontext context() const noexcept
+  {
+    return __ctx;
+  }
+
+  //! @brief Retrieve the device on which this logical device resides
+  [[nodiscard]] constexpr device_ref underlying_device() const noexcept
+  {
+    return __dev_id;
+  }
+
+  //! @brief Retrieve the kind of logical device stored in this object
+  //! The kind indicates if this logical_device holds a device or green_context
+  [[nodiscard]] constexpr kinds kind() const noexcept
+  {
+    return __kind;
+  }
+
+  //! @brief Construct logical_device from a device ordinal
+  //!
+  //! Constructing a logical_device for a given device ordinal has a side effect of initializing that device
+  explicit logical_device(int __id)
+      : __dev_id(__id)
+      , __kind(kinds::device)
+      , __ctx(::cuda::__physical_devices()[__id].__primary_context())
+  {}
+
+  //! @brief Construct logical_device from a device_ref
+  //!
+  //! Constructing a logical_device for a given device_ref has a side effect of initializing that device
+  explicit logical_device(device_ref __dev)
+      : logical_device(__dev.get())
+  {}
+
+#if _CCCL_CTK_AT_LEAST(12, 5)
+  //! @brief Construct logical_device from a green_context
+  logical_device(const green_context& __gctx)
+      : __dev_id(__gctx.__dev_id)
+      , __kind(kinds::green_context)
+      , __ctx(__gctx.__transformed)
+  {}
+#endif // _CCCL_CTK_AT_LEAST(12, 5)
+
+  //! @brief Compares two logical_devices for equality
+  //!
+  //! @param __lhs The first `logical_device` to compare
+  //! @param __rhs The second `logical_device` to compare
+  //! @return true if `lhs` and `rhs` refer to the same logical device
+  [[nodiscard]] friend bool operator==(logical_device __lhs, logical_device __rhs) noexcept
+  {
+    return __lhs.__ctx == __rhs.__ctx;
+  }
+
+#if _CCCL_STD_VER <= 2017
+  //! @brief Compares two logical_devices for inequality
+  //!
+  //! @param __lhs The first `logical_device` to compare
+  //! @param __rhs The second `logical_device` to compare
+  //! @return true if `lhs` and `rhs` refer to the different logical device
+  [[nodiscard]] friend bool operator!=(logical_device __lhs, logical_device __rhs) noexcept
+  {
+    return __lhs.__ctx != __rhs.__ctx;
+  }
+#endif // _CCCL_STD_VER <= 2017
+
+private:
+  friend __logical_device_access;
+  // This might be a CUdevice as well
+  int __dev_id = 0;
+  kinds __kind;
+  CUcontext __ctx = nullptr;
+
+  logical_device(int __id, CUcontext __context, kinds __k)
+      : __dev_id(__id)
+      , __kind(__k)
+      , __ctx(__context)
+  {}
+};
+
+struct __logical_device_access
+{
+  static logical_device make_logical_device(int __id, CUcontext __context, logical_device::kinds __k)
+  {
+    return logical_device(__id, __context, __k);
+  }
+};
+} // namespace cuda::experimental
+
+#include <cuda/std/__cccl/epilogue.h>
+
+#endif // _CUDAX__DEVICE_LOGICAL_DEVICE_CUH