[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/README.md

@@ -1,32 +0,0 @@
-## CUDA Experimental: Library for experimental features in CUDA Core Compute Libraries.
-CUDA Experimental serves as a distribution channel for features that are considered experimental in the CUDA Core Compute Libraries.
-Some of them are still actively designed or developed and their API is evolving.
-Some of them are specific to one hardware architecture and are still looking for a generic and forward compatible exposure.
-Finally, some of them need to prove useful enough to deserve long term support.
-
-**All APIs available in CUDA Experimental are not considered stable and can change without a notice.** They can also be deprecated or removed on a much faster cadence than in other CCCL libraries.
-
-Features are exposed here for the CUDA C++ community to experiment with and provide feedback on how to shape it to best fit their use cases.
-Once we become confident a feature is ready and would be a great permanent addition in CCCL, it will become a part of some other CCCL library with a stable API.
-
-## Installation
-CUDA Experimental library is **not** distributed with the CUDA Toolkit like the rest of CCCL. It is only available on the [CCCL GitHub repository](https://github.com/NVIDIA/cccl).
-
-CUDA Experimental compilation requires C++17 standard or newer. Supported compilers are:
-
-CUDA Compilers:
-- NVCC 12.3+
-
-NVCC host compilers:
-- GCC 7+
-- Clang 9+
-- MSVC 2019+
-
-Everything in CUDA Experimental is header-only, so cloning and including it in a simple project is as easy as the following:
-```bash
-git clone https://github.com/NVIDIA/cccl.git
-# Note:
-nvcc -Icccl/cudax/include main.cu -o main
-```
-
-A CMake target `cudax::cudax` is available as part of the CCCL package when `CCCL_ENABLE_UNSTABLE` is set to a truthy value before calling `find_package` or `add_subdirectory`.