[CCCL] 瘦身 + 补全: 移除 cudax/python/libcudacxx-tests 冗余文件, 新增 c2h 测试助手 + cmake 构建系统 + 8 个 CUDA thrust examples

变更摘要:
- 删除: cudax/ (783 files, 7.2M) — 实验性组件，竞赛不需要
- 删除: python/ (226 files, 2.0M) — Python 绑定，竞赛不需要
- 删除: libcudacxx/{test,benchmarks,codegen,cmake,share} (4432 files, 31M)
  保留: libcudacxx/include/ (1463 headers, cuda::std 编译依赖)
- 新增: c2h/ (27 files) — CUB Catch2 测试辅助头文件，编译 243 个测试必需
- 新增: cmake/ (29 files) — CCCL 原生 CMake 构建系统
- 新增: thrust/examples/cuda/ (7 files) + cpp_integration/ (1 file)
  async_reduce, custom_temporary_allocation, explicit_cuda_stream,
  global_device_vector, range_view, unwrap_pointer, wrap_pointer, device

结果: cccl_upstream 从 74M→35M (瘦身 53%), 核心内容 100% 保留:
  27/27 tuning headers, 78 benchmarks, 243 tests,
  60 thrust examples, 18 CUB examples, 全部编译头文件

cccl_upstream/cudax/test/coop/shuffle/threads_within_warp.cu

@@ -1,141 +0,0 @@
-//===----------------------------------------------------------------------===//
-//
-// 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) 2026 NVIDIA CORPORATION & AFFILIATES.
-//
-//===----------------------------------------------------------------------===//
-
-#include <cuda/devices>
-#include <cuda/hierarchy>
-#include <cuda/launch>
-#include <cuda/stream>
-#include <cuda/type_traits>
-
-#include <cuda/experimental/coop.cuh>
-#include <cuda/experimental/group.cuh>
-
-#include "testing.cuh"
-
-template <class T>
-__device__ T make_instance_for(unsigned rank)
-{
-  if constexpr (cuda::is_vector_type_v<T>)
-  {
-    using U = cuda::scalar_type_t<T>;
-    return T{static_cast<U>(rank), static_cast<U>(rank), static_cast<U>(rank)};
-  }
-  else
-  {
-    return static_cast<T>(rank);
-  }
-}
-
-template <class T, class Group>
-__device__ void test_group(const Group& group)
-{
-  // Exit all threads that are not part of the group.
-  if (!cuda::gpu_thread.is_part_of(group))
-  {
-    return;
-  }
-
-  const auto my_rank  = cuda::gpu_thread.rank_as<unsigned>(group);
-  const auto my_value = make_instance_for<T>(my_rank);
-
-  // Test identity.
-  REQUIRE(cudax::coop::shuffle(group, my_value, my_rank) == my_value);
-
-  // Test broadcast from root rank.
-  REQUIRE(cudax::coop::shuffle(group, my_value, 0) == make_instance_for<T>(0));
-
-  // Test broadcast from last rank.
-  {
-    const auto other_rank = cuda::gpu_thread.count_as<unsigned>(group) - 1;
-    REQUIRE(cudax::coop::shuffle(group, my_value, other_rank) == make_instance_for<T>(other_rank));
-  }
-
-  // Test my_rank + 1.
-  {
-    const auto other_rank = (my_rank + 1) % cuda::gpu_thread.count_as<unsigned>(group);
-    REQUIRE(cudax::coop::shuffle(group, my_value, other_rank) == make_instance_for<T>(other_rank));
-  }
-
-  // Test my_rank + 6.
-  {
-    const auto other_rank = (my_rank + 6) % cuda::gpu_thread.count_as<unsigned>(group);
-    REQUIRE(cudax::coop::shuffle(group, my_value, other_rank) == make_instance_for<T>(other_rank));
-  }
-
-  // Test my_rank +- 1 based on whether my_rank is even or not.
-  {
-    const auto other_rank =
-      ((my_rank % 2 == 0) ? my_rank + 1 : my_rank - 1) % cuda::gpu_thread.count_as<unsigned>(group);
-    REQUIRE(cudax::coop::shuffle(group, my_value, other_rank) == make_instance_for<T>(other_rank));
-  }
-
-  // Test my_rank +- 3 based on whether my_rank is even or not.
-  {
-    const auto other_rank =
-      ((my_rank % 2 == 0) ? my_rank + 3 : my_rank - 3) % cuda::gpu_thread.count_as<unsigned>(group);
-    REQUIRE(cudax::coop::shuffle(group, my_value, other_rank) == make_instance_for<T>(other_rank));
-  }
-}
-
-struct CustomBinaryPartition
-{
-  template <class MappingResult>
-  __device__ bool operator()(MappingResult mapping_result)
-  {
-    switch (mapping_result.unit_rank())
-    {
-      case 1:
-      case 5:
-      case 14:
-      case 15:
-      case 31:
-        return true;
-      default:
-        return false;
-    }
-  }
-};
-
-template <class T, class Config>
-__device__ void test_type(const Config& config)
-{
-  const cudax::this_warp warp{config};
-  test_group<T>(cudax::group{cuda::gpu_thread, warp, cudax::identity_mapping{}, cudax::lane_synchronizer{}});
-  test_group<T>(cudax::group{cuda::gpu_thread, warp, cudax::group_by<4>{}, cudax::lane_synchronizer{}});
-  test_group<T>(cudax::group{cuda::gpu_thread, warp, cudax::group_by{1}, cudax::lane_synchronizer{}});
-  test_group<T>(
-    cudax::group{cuda::gpu_thread, warp, cudax::group_by{3, cudax::non_exhaustive}, cudax::lane_synchronizer{}});
-  test_group<T>(
-    cudax::group{cuda::gpu_thread, warp, cudax::binary_partition{CustomBinaryPartition{}}, cudax::lane_synchronizer{}});
-}
-
-struct TestKernel
-{
-  template <class Config>
-  __device__ void operator()(const Config& config)
-  {
-    test_type<signed char>(config);
-    test_type<unsigned>(config);
-    test_type<unsigned long long>(config);
-    test_type<longlong3>(config);
-  }
-};
-
-C2H_TEST("shuffle/threads_within_warp", "[shuffle][threads_within_warp]")
-{
-  const auto device = cuda::devices[0];
-
-  const cuda::stream stream{device};
-
-  const auto config = cuda::make_config(cuda::grid_dims<1>(), cuda::block_dims<32>());
-  cuda::launch(stream, config, TestKernel{});
-
-  stream.sync();
-}