[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/cmake/header_test.in.cu

@@ -1,66 +0,0 @@
-// This source file checks that:
-// 1) Header <@header@> compiles without error.
-// 2) Common macro collisions with platform/system headers are avoided.
-// 3) half/bf16 aren't included when these are explicitly disabled.
-
-// Define CUDAX_MACRO_CHECK(macro, header), which emits a diagnostic indicating
-// a potential macro collision and halts.
-//
-// Use raw platform checks instead of the CCCL macros since we
-// don't want to #include any headers other than the one being tested.
-//
-// This is only implemented for MSVC/GCC/Clang.
-#if defined(_MSC_VER) // MSVC
-
-// Fake up an error for MSVC
-#  define CUDAX_MACRO_CHECK_IMPL(msg)             \
-    /* Print message that looks like an error: */ \
-    __pragma(message(__FILE__ ":" CUDAX_MACRO_CHECK_IMPL0(__LINE__) ": error: " #msg)) static_assert(false, #msg);
-#  define CUDAX_MACRO_CHECK_IMPL0(x) CUDAX_MACRO_CHECK_IMPL1(x)
-#  define CUDAX_MACRO_CHECK_IMPL1(x) #x
-
-#elif defined(__clang__) || defined(__GNUC__)
-
-// GCC/clang are easy:
-#  define CUDAX_MACRO_CHECK_IMPL(msg)   CUDAX_MACRO_CHECK_IMPL0(GCC error #msg)
-#  define CUDAX_MACRO_CHECK_IMPL0(expr) _Pragma(#expr)
-
-#endif
-
-// Hacky way to build a string, but it works on all tested platforms.
-#define CUDAX_MACRO_CHECK(MACRO, HEADER) \
-  CUDAX_MACRO_CHECK_IMPL(Identifier MACRO should not be used from CCCL headers due to conflicts with HEADER macros.)
-
-// complex.h conflicts
-#define I CUDAX_MACRO_CHECK('I', complex.h)
-
-// windows.h conflicts
-// @eniebler 2024-08-30: This test is disabled because it causes build
-// failures in some configurations.
-// #define small CUDAX_MACRO_CHECK('small', windows.h)
-// We can't enable these checks without breaking some builds -- some standard
-// library implementations unconditionally `#undef` these macros, which then
-// causes random failures later.
-// Leaving these commented out as a warning: Here be dragons.
-// #define min(...) CUDAX_MACRO_CHECK('min', windows.h)
-// #define max(...) CUDAX_MACRO_CHECK('max', windows.h)
-
-// termios.h conflicts (NVIDIA/thrust#1547)
-#define B0 CUDAX_MACRO_CHECK("B0", termios.h)
-
-#include <@header@>
-
-#if defined(CCCL_DISABLE_BF16_SUPPORT)
-#  if defined(__CUDA_BF16_TYPES_EXIST__)
-#    error We should not include cuda_bf16.h when BF16 support is disabled
-#  endif // __CUDA_BF16_TYPES_EXIST__
-#endif // CCCL_DISABLE_BF16_SUPPORT
-
-#if defined(CCCL_DISABLE_FP16_SUPPORT)
-#  if defined(__CUDA_FP16_TYPES_EXIST__)
-#    error We should not include cuda_fp16.h when half support is disabled
-#  endif // __CUDA_FP16_TYPES_EXIST__
-#  if defined(__CUDA_BF16_TYPES_EXIST__)
-#    error We should not include cuda_bf16.h when half support is disabled
-#  endif // __CUDA_BF16_TYPES_EXIST__
-#endif // CCCL_DISABLE_FP16_SUPPORT