[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/thrust/testing/unittest/cuda/testframework.h

@@ -0,0 +1,25 @@
+#pragma once
+
+#include <thrust/system/cuda/memory.h>
+#include <thrust/system_error.h>
+
+#include <vector>
+
+#include <unittest/testframework.h>
+
+class CUDATestDriver : public UnitTestDriver
+{
+public:
+  int current_device_architecture() const;
+
+private:
+  std::vector<int> target_devices(const ArgumentMap& kwargs);
+
+  bool check_cuda_error(bool concise);
+
+  bool post_test_smoke_check(const UnitTest& test, bool concise) override;
+
+  bool run_tests(const ArgumentSet& args, const ArgumentMap& kwargs) override;
+};
+
+UnitTestDriver& driver_instance(thrust::system::cuda::tag);