[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/examples/stf/08-cub-reduce.cu

@@ -1,95 +0,0 @@
-//===----------------------------------------------------------------------===//
-//
-// Part of CUDASTF 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) 2022-2024 NVIDIA CORPORATION & AFFILIATES.
-//
-//===----------------------------------------------------------------------===//
-
-/**
- * @file
- * @brief Example of reduction implementing using CUB kernels
- */
-
-#include <thrust/device_vector.h>
-
-#include <cuda/experimental/stf.cuh>
-
-using namespace cuda::experimental::stf;
-
-template <int BLOCK_THREADS, typename T>
-__global__ void reduce(slice<const T> values, slice<T> partials, size_t nelems)
-{
-  using namespace cub;
-  typedef BlockReduce<T, BLOCK_THREADS> BlockReduceT;
-
-  auto thread_id = BLOCK_THREADS * blockIdx.x + threadIdx.x;
-
-  // Local reduction
-  T local_sum = 0;
-  for (size_t ind = thread_id; ind < nelems; ind += blockDim.x * gridDim.x)
-  {
-    local_sum += values(ind);
-  }
-
-  __shared__ typename BlockReduceT::TempStorage temp_storage;
-
-  // Per-thread tile data
-  T result = BlockReduceT(temp_storage).Sum(local_sum);
-
-  if (threadIdx.x == 0)
-  {
-    partials(blockIdx.x) = result;
-  }
-}
-
-template <typename Ctx>
-void run()
-{
-  Ctx ctx;
-
-  const size_t N          = 1024 * 16;
-  const size_t BLOCK_SIZE = 128;
-  const size_t num_blocks = 32;
-
-  int *X, ref_tot;
-
-  X       = new int[N];
-  ref_tot = 0;
-
-  for (size_t ind = 0; ind < N; ind++)
-  {
-    X[ind] = rand() % N;
-    ref_tot += X[ind];
-  }
-
-  auto values   = ctx.logical_data(X, {N});
-  auto partials = ctx.logical_data(shape_of<slice<int>>(num_blocks));
-  auto result   = ctx.logical_data(shape_of<slice<int>>(1));
-
-  ctx.task(values.read(), partials.write(), result.write())->*[&](auto stream, auto values, auto partials, auto result) {
-    // reduce values into partials
-    reduce<BLOCK_SIZE, int><<<num_blocks, BLOCK_SIZE, 0, stream>>>(values, partials, N);
-
-    // reduce partials on a single block into result
-    reduce<BLOCK_SIZE, int><<<1, BLOCK_SIZE, 0, stream>>>(partials, result, num_blocks);
-  };
-
-  ctx.host_launch(result.read())->*[&](auto p) {
-    if (p(0) != ref_tot)
-    {
-      fprintf(stderr, "INCORRECT RESULT: p sum = %d, ref tot = %d\n", p(0), ref_tot);
-      abort();
-    }
-  };
-
-  ctx.finalize();
-}
-
-int main()
-{
-  run<stream_ctx>();
-  run<graph_ctx>();
-}