[CCCL] Add missing CCCL components: c2h, nvbench_helper, cmake, cudax, AGENTS.md

Added 863 files from NVIDIA/cccl sparse checkout:
- c2h/ (27 files): Catch2 test helpers — generators, validators, runner
- nvbench_helper/ (10 files): Benchmark harness utilities
- cmake/ (29 files): CMake presets and build helpers
- cudax/ (794 files): Experimental CUDA extensions
- AGENTS.md: NVIDIA's official AI agent instructions for CCCL
- CMakePresets.json: Standardized build configurations
- cccl-version.json: Version tracking

Also added CCCL_ASSET_MAP.md mapping all 4295 CCCL files to
competition value and PRD items.

cccl_upstream now covers 100% of competition-critical assets:
- 27 tuning headers (SM80/90/100 benchmark data)
- 32 dispatch headers (algorithm implementations)
- 60 Thrust examples (correctness verification)
- 217 CUB Catch2 tests (regression matrix)
- 153 CUB benchmarks (parameter space search)
- 18 CUB examples (API verification)
- 27 test helpers + benchmark harness
- 794 cudax experimental extensions

cccl_upstream/cudax/examples/vector.cuh

@@ -0,0 +1,159 @@
+//===----------------------------------------------------------------------===//
+//
+// 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) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _CUDAX__CONTAINER_VECTOR
+#define _CUDAX__CONTAINER_VECTOR
+
+#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 <thrust/device_vector.h>
+#include <thrust/host_vector.h>
+
+#include <cuda/__stream/stream_ref.h>
+#include <cuda/std/__type_traits/maybe_const.h>
+#include <cuda/std/span>
+
+#include <cuda/experimental/__detail/utility.cuh>
+#include <cuda/experimental/__launch/param_kind.cuh>
+
+#include <cstdio>
+
+namespace cuda::experimental
+{
+using ::cuda::std::span;
+using ::thrust::device_vector;
+using ::thrust::host_vector;
+
+template <typename _Ty>
+class vector
+{
+public:
+  vector() = default;
+  explicit vector(size_t __n)
+      : __h_(__n)
+  {}
+
+  _Ty& operator[](size_t __i) noexcept
+  {
+    __dirty_ = true;
+    return __h_[__i];
+  }
+
+  const _Ty& operator[](size_t __i) const noexcept
+  {
+    return __h_[__i];
+  }
+
+private:
+  void sync_host_to_device([[maybe_unused]] ::cuda::stream_ref __str, __detail::__param_kind __p) const
+  {
+    if (__dirty_)
+    {
+      if (__p == __detail::__param_kind::_out)
+      {
+        // There's no need to copy the data from host to device if the data is
+        // only going to be written to. We can just allocate the device memory.
+        __d_.resize(__h_.size());
+      }
+      else
+      {
+        // TODO: use a memcpy async here
+        __d_ = __h_;
+      }
+      __dirty_ = false;
+    }
+  }
+
+  void sync_device_to_host(::cuda::stream_ref __str, __detail::__param_kind __p) const
+  {
+    if (__p != __detail::__param_kind::_in)
+    {
+      // TODO: use a memcpy async here
+      __str.sync(); // wait for the kernel to finish executing
+      __h_ = __d_;
+    }
+  }
+
+  template <__detail::__param_kind _Kind>
+  class __action //: private __detail::__immovable
+  {
+    using __cv_vector = ::cuda::std::__maybe_const<_Kind == __detail::__param_kind::_in, vector>;
+
+  public:
+    explicit __action(::cuda::stream_ref __str, __cv_vector& __v)
+        : __str_(__str)
+        , __v_(__v)
+    {
+      __v_.sync_host_to_device(__str_, _Kind);
+    }
+
+    __action(__action&&) = delete;
+
+    ~__action()
+    {
+      try
+      {
+        __v_.sync_device_to_host(__str_, _Kind);
+      }
+      catch (const ::std::exception& e)
+      {
+        static_cast<void>(
+          ::fprintf(stderr, "Exception occurred during host to device synchronization: %s\n", e.what()));
+      }
+      catch (...)
+      {
+        static_cast<void>(::fprintf(stderr, "Unknown exception occurred during host to device synchronization\n"));
+      }
+    }
+
+    ::cuda::std::span<_Ty> transformed_argument() const
+    {
+      return {__v_.__d_.data().get(), __v_.__d_.size()};
+    }
+
+  private:
+    ::cuda::stream_ref __str_;
+    __cv_vector& __v_;
+  };
+
+  [[nodiscard]] friend __action<__detail::__param_kind::_inout>
+  transform_launch_argument(::cuda::stream_ref __str, vector& __v)
+  {
+    return __action<__detail::__param_kind::_inout>{__str, __v};
+  }
+
+  [[nodiscard]] friend __action<__detail::__param_kind::_in>
+  transform_launch_argument(::cuda::stream_ref __str, const vector& __v)
+  {
+    return __action<__detail::__param_kind::_in>{__str, __v};
+  }
+
+  template <__detail::__param_kind _Kind>
+  [[nodiscard]] friend __action<_Kind>
+  transform_launch_argument(::cuda::stream_ref __str, __detail::__box<vector, _Kind> __b)
+  {
+    return __action<_Kind>{__str, __b.__val};
+  }
+
+  mutable host_vector<_Ty> __h_;
+  mutable device_vector<_Ty> __d_{};
+  mutable bool __dirty_ = true;
+};
+} // namespace cuda::experimental
+
+#endif