[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/cub/test/internal/catch2_test_fast_div_mod.cu

@@ -0,0 +1,55 @@
+// SPDX-FileCopyrightText: Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3-Clause
+#include <cub/config.cuh>
+
+#include "c2h/catch2_test_helper.h"
+#include "c2h/utility.h"
+
+/***********************************************************************************************************************
+ * TEST CASES
+ **********************************************************************************************************************/
+
+using index_types =
+  c2h::type_list<int8_t,
+                 uint8_t,
+                 int16_t,
+                 uint16_t,
+                 int32_t,
+                 uint32_t
+#if TEST_INT128()
+                 ,
+                 int64_t,
+                 uint64_t
+#endif
+                 >;
+
+C2H_TEST("FastDivMod random", "[FastDivMod][Random]", index_types)
+{
+  using cub::detail::fast_div_mod;
+  using index_type         = c2h::get<0, TestType>;
+  constexpr auto max_value = +cuda::std::numeric_limits<index_type>::max();
+  auto dividend            = GENERATE_COPY(take(20, random(+index_type{1}, max_value)));
+  auto divisor             = GENERATE_COPY(take(20, random(+index_type{1}, max_value)));
+  fast_div_mod<index_type> div_mod(static_cast<index_type>(divisor));
+  CAPTURE(c2h::type_name<index_type>(), dividend, divisor);
+  static_assert(std::is_same_v<decltype(dividend / divisor), decltype(div_mod(dividend).quotient)>,
+                "quotient type mismatch");
+  REQUIRE(dividend / divisor == div_mod(dividend).quotient);
+  REQUIRE(dividend % divisor == div_mod(dividend).remainder);
+}
+
+C2H_TEST("FastDivMod edge cases", "[FastDivMod][EdgeCases]", index_types)
+{
+  using cub::detail::fast_div_mod;
+  using index_type         = c2h::get<0, TestType>;
+  constexpr auto max_value = cuda::std::numeric_limits<index_type>::max();
+  CAPTURE(c2h::type_name<index_type>());
+  // divisor/dividend == max
+  fast_div_mod<index_type> div_mod_max(max_value);
+  REQUIRE(1 == div_mod_max(max_value).quotient);
+  REQUIRE(0 == div_mod_max(max_value).remainder);
+  // divisor == 10, dividend == 0
+  fast_div_mod<index_type> div_mod_min(10);
+  REQUIRE(0 == div_mod_min(0).quotient);
+  REQUIRE(0 == div_mod_min(0).remainder);
+}