[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/03-temporary-data.cu

@@ -1,69 +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 This example illustrates how we can create temporary data from shapes, and use them in tasks
- */
-
-#include <cuda/experimental/stf.cuh>
-
-using namespace cuda::experimental::stf;
-
-int main()
-{
-  const int n = 4096;
-  int X[n];
-  int Y[n];
-
-  for (size_t i = 0; i < n; i++)
-  {
-    X[i] = 3 * i;
-    Y[i] = 2 * i - 3;
-  }
-
-  context ctx;
-
-  auto lX = ctx.logical_data(X);
-  auto lY = ctx.logical_data(Y);
-
-  // Select an odd number
-  int niter = 19;
-  assert(niter % 2 == 1);
-
-  for (int iter = 0; iter < niter; iter++)
-  {
-    // We here define a temporary vector with the same shape as X, for which there is no existing copy
-    // This data handle has a limited scope, so that it is automatically destroyed at each iteration of the loop
-    auto tmp = ctx.logical_data(lX.shape());
-
-    ctx.task(lY.rw(), lX.rw(), tmp.write())->*[](cudaStream_t s, auto sY, auto sX, auto sTMP) {
-      // We swap X and Y using TMP as temporary buffer
-      // TMP = X
-      cuda_safe_call(
-        cudaMemcpyAsync(sTMP.data_handle(), sX.data_handle(), n * sizeof(int), cudaMemcpyDeviceToDevice, s));
-      // X = Y
-      cuda_safe_call(cudaMemcpyAsync(sX.data_handle(), sY.data_handle(), n * sizeof(int), cudaMemcpyDeviceToDevice, s));
-      // Y = TMP
-      cuda_safe_call(
-        cudaMemcpyAsync(sY.data_handle(), sTMP.data_handle(), n * sizeof(int), cudaMemcpyDeviceToDevice, s));
-    };
-  }
-
-  ctx.finalize();
-
-  // We have exchanged an odd number of times, so they must be inverted
-  for (size_t i = 0; i < n; i++)
-  {
-    assert(X[i] == 2 * i - 3);
-    assert(Y[i] == 3 * i);
-  }
-}