[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/namespace_wrapped.cu

@@ -0,0 +1,38 @@
+// Wrap thrust and cub in different enclosing namespaces
+// (In practice, you probably want these to be the same, in which case just
+// set THRUST_CUB_WRAPPED_NAMESPACE to set both).
+#define THRUST_WRAPPED_NAMESPACE wrap_thrust
+#define CUB_WRAPPED_NAMESPACE    wrap_cub
+#define CCCL_IGNORE_DEPRECATED_API
+
+#include <thrust/device_vector.h>
+#include <thrust/host_vector.h>
+#include <thrust/transform.h>
+
+#include <cuda/iterator>
+
+#include <unittest/unittest.h>
+
+// Test that we can use a few common utilities and algorithms from a wrapped
+// namespace at runtime. More extensive testing is performed by the header
+// tests and the check_namespace.cmake test.
+void TestWrappedNamespace()
+{
+  const std::size_t n = 2048;
+
+  const auto in_1_begin = ::cuda::make_constant_iterator<int>(12);
+  const auto in_2_begin = ::cuda::make_counting_iterator<int>(1024);
+
+  // Check that the qualifier resolves properly:
+  THRUST_NS_QUALIFIER::device_vector<int> d_out(n);
+
+  ::wrap_thrust::thrust::transform(in_1_begin, in_1_begin + n, in_2_begin, d_out.begin(), ::cuda::std::plus<>{});
+
+  ::wrap_thrust::thrust::host_vector<int> h_out(d_out);
+
+  for (std::size_t i = 0; i < n; ++i)
+  {
+    ASSERT_EQUAL(h_out[i], static_cast<int>(i) + 1024 + 12);
+  }
+}
+DECLARE_UNITTEST(TestWrappedNamespace);