[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/caching_allocator.cu

@@ -0,0 +1,24 @@
+#include <thrust/detail/config.h>
+
+#include <thrust/detail/caching_allocator.h>
+
+#include <unittest/unittest.h>
+
+template <typename Allocator>
+void test_implementation(Allocator alloc)
+{
+  using Traits = typename cuda::std::allocator_traits<Allocator>;
+  using Ptr    = typename Allocator::pointer;
+
+  Ptr p = Traits::allocate(alloc, 123);
+  Traits::deallocate(alloc, p, 123);
+
+  Ptr p2 = Traits::allocate(alloc, 123);
+  ASSERT_EQUAL(p, p2);
+}
+
+void TestSingleDeviceTLSCachingAllocator()
+{
+  test_implementation(thrust::detail::single_device_tls_caching_allocator());
+};
+DECLARE_UNITTEST(TestSingleDeviceTLSCachingAllocator);