[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/group/segmented_algorithm.cu

@@ -1,113 +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 <cub/block/block_reduce.cuh>
-
-#include <cuda/buffer>
-#include <cuda/devices>
-#include <cuda/hierarchy>
-#include <cuda/iterator>
-#include <cuda/launch>
-#include <cuda/std/algorithm>
-#include <cuda/std/cstddef>
-#include <cuda/std/execution>
-#include <cuda/std/optional>
-#include <cuda/std/span>
-#include <cuda/std/type_traits>
-#include <cuda/std/utility>
-#include <cuda/stream>
-
-#include <cuda/experimental/group.cuh>
-
-#include "group_testing.cuh"
-
-namespace
-{
-struct DeviceSegmentedSumKernel
-{
-  template <class Config, class T, cuda::std::size_t SegmentSize, class GroupFn, class UnitFn>
-  __device__ void operator()(
-    Config config,
-    const T* in,
-    T* out,
-    cuda::std::size_t nsegments,
-    cuda::std::integral_constant<cuda::std::size_t, SegmentSize>,
-    GroupFn group_fn,
-    UnitFn unit_fn)
-  {
-    constexpr auto nitems_per_thread = SegmentSize / cuda::gpu_thread.static_count(cuda::block, config);
-
-    const auto segment_offset = SegmentSize * cuda::block.rank(cuda::grid, config);
-
-    T items[nitems_per_thread];
-    for (cuda::std::size_t i = 0; i < nitems_per_thread; ++i)
-    {
-      const auto offset = cuda::gpu_thread.rank(cuda::block, config) + i * cuda::gpu_thread.count(cuda::block, config);
-      items[i]          = *(in + segment_offset + offset);
-    }
-
-    group_fn(cudax::this_block{config}, cuda::std::span{items});
-
-    using BlockReduce = cub::BlockReduce<T, static_cast<int>(cuda::gpu_thread.static_count(cuda::block, config))>;
-    __shared__ typename BlockReduce::TempStorage scratch;
-
-    const auto result = BlockReduce{scratch}.Sum(items);
-    if (cuda::gpu_thread.rank(cuda::block, config) == 0)
-    {
-      out[cuda::block.rank(cuda::grid, config)] = unit_fn(result);
-    }
-  }
-};
-
-template <class T, cuda::std::size_t SegmentSize, class GroupFn, class UnitFn>
-void device_segmented_sum(
-  cuda::stream_ref stream,
-  const T* in,
-  T* out,
-  cuda::std::size_t nsegments,
-  cuda::std::integral_constant<cuda::std::size_t, SegmentSize> segment_size,
-  GroupFn group_fn,
-  UnitFn unit_fn)
-{
-  const auto config =
-    cuda::make_config(cuda::grid_dims(dim3{static_cast<unsigned>(nsegments)}), cuda::block_dims<SegmentSize>());
-  cuda::launch(stream, config, DeviceSegmentedSumKernel{}, in, out, nsegments, segment_size, group_fn, unit_fn);
-}
-} // namespace
-
-C2H_TEST("Segmented algorithm", "[group]")
-{
-  const auto device = cuda::devices[0];
-
-  const cuda::stream stream{device};
-
-  auto in  = cuda::make_device_buffer<int>(stream, device, 1024, 1);
-  auto out = cuda::make_device_buffer<int>(stream, device, 8, cuda::no_init);
-
-  device_segmented_sum(
-    stream,
-    in.data(),
-    out.data(),
-    8,
-    cuda::std::integral_constant<cuda::std::size_t, 128>{},
-    [] __device__(auto group, auto items) {
-      for (auto& item : items)
-      {
-        item *= 2;
-      }
-      group.sync();
-    },
-    [] __device__(auto value) {
-      return value / 2;
-    });
-  stream.sync();
-
-  CHECK(cuda::std::equal(cuda::execution::gpu, out.begin(), out.end(), cuda::constant_iterator{128}));
-}