[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/stf/allocators/adapter.cu

@@ -1,60 +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.
-//
-//===----------------------------------------------------------------------===//
-
-#include <cuda/experimental/__stf/allocators/adapters.cuh>
-#include <cuda/experimental/stf.cuh>
-
-using namespace cuda::experimental::stf;
-
-int main()
-{
-  double* d_ptrA;
-  const size_t N     = 128 * 1024;
-  const size_t NITER = 10;
-
-  // User allocated memory
-  cuda_safe_call(cudaMalloc(&d_ptrA, N * sizeof(double)));
-
-  async_resources_handle handle;
-
-  cudaStream_t stream;
-  cuda_safe_call(cudaStreamCreate(&stream));
-
-  for (size_t i = 0; i < NITER; i++)
-  {
-    graph_ctx ctx(stream, handle);
-
-    // The uncached allocator of the context will be using cudaMallocAsync(...,
-    // stream) to avoid creating memory nodes in the graph (because they are
-    // costly and caching the graph also keeps memory allocated)
-    auto wrapper = stream_adapter(ctx, stream);
-
-    ctx.set_allocator(block_allocator<buddy_allocator>(ctx, wrapper.allocator()));
-
-    auto A = ctx.logical_data(make_slice(d_ptrA, N), data_place::current_device());
-
-    for (size_t k = 0; k < 4; k++)
-    {
-      auto tmp  = ctx.logical_data(A.shape());
-      auto tmp2 = ctx.logical_data(A.shape());
-      // Test device and managed memory
-      ctx.parallel_for(A.shape(), A.read(), tmp.write(), tmp2.write(data_place::managed()))
-          ->*[] __device__(size_t i, auto a, auto tmp, auto tmp2) {
-                tmp(i)  = a(i);
-                tmp2(i) = a(i);
-              };
-    }
-
-    ctx.finalize();
-
-    wrapper.clear();
-  }
-  cuda_safe_call(cudaStreamSynchronize(stream));
-}