[INFRA] Import NVIDIA/CCCL upstream as optimization reference library

CCCL (CUDA C++ Core Libraries) provides:
- CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk)
- Thrust: high-level parallel algorithms (transform_reduce, sort, scan)
- libcudacxx: CUDA C++ standard library (atomics, barriers, memory)
- cudax: experimental features (memory resources, allocators)
- Tuning policies: per-SM hardware-specific algorithm parameters

Competition optimization vectors mapped to CCCL:
- Output TPS (83% weight): warp_reduce, block_reduce, device_topk
- Input TPS (14% weight): device_scan, block_load, prefetch
- Cache TPS (3% weight): prefix caching strategy patterns
- Memory (0.9 util): pooled/cached/buddy allocators

Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only)
License: Apache-2.0

cccl_upstream/libcudacxx/test/support/assert_checkpoint.h

@@ -0,0 +1,97 @@
+#ifndef SUPPORT_ASSERT_CHECKPOINT_H
+#define SUPPORT_ASSERT_CHECKPOINT_H
+
+#include <csignal>
+#include <cstdlib>
+#include <iostream>
+
+struct Checkpoint
+{
+  const char* file;
+  const char* func;
+  int line;
+  const char* msg;
+
+  Checkpoint()
+      : file(nullptr)
+      , func(nullptr)
+      , line(-1)
+      , msg(nullptr)
+  {}
+  Checkpoint(const char* xfile, const char* xfunc, int xline, const char* xmsg)
+      : file(xfile)
+      , func(xfunc)
+      , line(xline)
+      , msg(xmsg)
+  {}
+
+  template <class Stream>
+  void print(Stream& s) const
+  {
+    if (!file)
+    {
+      s << "NO CHECKPOINT\n";
+      return;
+    }
+    s << file << ":" << line << " " << func << ": Checkpoint";
+    if (msg)
+    {
+      s << " '" << msg << "'";
+    }
+    s << '\n';
+  }
+};
+
+inline Checkpoint& globalCheckpoint()
+{
+  static Checkpoint C;
+  return C;
+}
+
+inline void clearCheckpoint()
+{
+  globalCheckpoint() = Checkpoint();
+}
+
+#if defined(__GNUC__)
+#  define CHECKPOINT_FUNCTION_NAME __PRETTY_FUNCTION__
+#else
+#  define CHECKPOINT_FUNCTION_NAME __func__
+#endif
+
+#define CHECKPOINT(msg) globalCheckpoint() = Checkpoint(__FILE__, CHECKPOINT_FUNCTION_NAME, __LINE__, msg);
+
+inline void checkpointSignalHandler(int signal)
+{
+  if (signal == SIGABRT)
+  {
+    globalCheckpoint().print(std::cerr);
+  }
+  else
+  {
+    std::cerr << "Unexpected signal " << signal << " received\n";
+  }
+  std::_Exit(EXIT_FAILURE);
+}
+
+inline bool initCheckpointHandler()
+{
+  using HandlerT     = void (*)(int);
+  static bool isInit = false;
+  if (isInit)
+  {
+    return true;
+  }
+  HandlerT prev_h = std::signal(SIGABRT, checkpointSignalHandler);
+  if (prev_h == SIG_ERR)
+  {
+    std::cerr << "Setup failed.\n";
+    std::_Exit(EXIT_FAILURE);
+  }
+  isInit = true;
+  return false;
+}
+
+static bool initDummy = initCheckpointHandler();
+
+#endif