[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/c/parallel.v2/test/freestanding/test_compiler.cpp

@@ -0,0 +1,66 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of CUDA Experimental 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) 2026 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#include <cassert>
+#include <cstdio>
+
+#include <cuda_runtime.h>
+
+#include <hostjit/config.hpp>
+#include <hostjit/jit_compiler.hpp>
+
+static const char* k_source = R"(
+#include <cuda_runtime.h>
+#include <cuda/std/version>
+
+__global__ void device_kernel(int* ptr)
+{
+  *ptr = 42;
+}
+
+extern "C" _CCCL_VISIBILITY_EXPORT void host_entry(int* ptr)
+{
+  device_kernel<<<1, 1>>>(ptr);
+}
+)";
+
+int main()
+{
+  // Detect Clang/CUDA configuration from the build environment
+  auto config = hostjit::detectDefaultConfig();
+
+  hostjit::JITCompiler compiler(config);
+  if (!compiler.compile(k_source))
+  {
+    std::fprintf(stderr, "HostJIT compilation failed:\n%s\n", compiler.getLastError().c_str());
+    return 1;
+  }
+
+  auto host_fn = compiler.getFunction<void (*)(int*)>("host_entry");
+  if (!host_fn)
+  {
+    std::fprintf(stderr, "Symbol 'host_entry' not found\n");
+    return 1;
+  }
+
+  int* d_ptr = nullptr;
+  cudaMalloc(&d_ptr, sizeof(int));
+
+  host_fn(d_ptr);
+  cudaDeviceSynchronize();
+
+  int result = 0;
+  cudaMemcpy(&result, d_ptr, sizeof(int), cudaMemcpyDeviceToHost);
+  cudaFree(d_ptr);
+
+  assert(result == 42 && "device kernel did not write expected value");
+  std::printf("freestanding compiler test passed (result=%d)\n", result);
+  return 0;
+}