[CCCL] Add missing CCCL components: c2h, nvbench_helper, cmake, cudax, AGENTS.md

Added 863 files from NVIDIA/cccl sparse checkout:
- c2h/ (27 files): Catch2 test helpers — generators, validators, runner
- nvbench_helper/ (10 files): Benchmark harness utilities
- cmake/ (29 files): CMake presets and build helpers
- cudax/ (794 files): Experimental CUDA extensions
- AGENTS.md: NVIDIA's official AI agent instructions for CCCL
- CMakePresets.json: Standardized build configurations
- cccl-version.json: Version tracking

Also added CCCL_ASSET_MAP.md mapping all 4295 CCCL files to
competition value and PRD items.

cccl_upstream now covers 100% of competition-critical assets:
- 27 tuning headers (SM80/90/100 benchmark data)
- 32 dispatch headers (algorithm implementations)
- 60 Thrust examples (correctness verification)
- 217 CUB Catch2 tests (regression matrix)
- 153 CUB benchmarks (parameter space search)
- 18 CUB examples (API verification)
- 27 test helpers + benchmark harness
- 794 cudax experimental extensions

cccl_upstream/cudax/test/stf/graph/concurrency_test.cu

@@ -0,0 +1,86 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of CUDASTF 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) 2022-2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#include <cuda/experimental/__stf/graph/graph_ctx.cuh>
+
+#include <iostream>
+
+/*
+ * The goal of this test is to ensure that using read access modes actually
+ * results in concurrent tasks
+ */
+
+using namespace cuda::experimental::stf;
+
+static __global__ void cuda_sleep_kernel(long long int clock_cnt)
+{
+  long long int start_clock  = clock64();
+  long long int clock_offset = 0;
+  while (clock_offset < clock_cnt)
+  {
+    clock_offset = clock64() - start_clock;
+  }
+}
+
+int main(int argc, char** argv)
+{
+  int NTASKS = 256;
+  int ms     = 40;
+
+  if (argc > 1)
+  {
+    NTASKS = atoi(argv[1]);
+  }
+
+  if (argc > 2)
+  {
+    ms = atoi(argv[2]);
+  }
+
+  // cudaDevAttrClockRate: Peak clock frequency in kilohertz;
+  int clock_rate;
+  cuda_safe_call(cudaDeviceGetAttribute(&clock_rate, cudaDevAttrClockRate, 0));
+  long long int clock_cnt = (long long int) (ms * clock_rate);
+
+  graph_ctx ctx;
+
+  int dummy[1];
+  auto handle = ctx.logical_data(dummy);
+
+  ctx.task(handle.rw())->*[](cudaGraph_t graph, auto /*unused*/) {
+    cudaGraphNode_t n;
+    cuda_safe_call(cudaGraphAddEmptyNode(&n, graph, nullptr, 0));
+  };
+
+  for (int iter = 0; iter < 10; iter++)
+  {
+    for (int k = 0; k < NTASKS; k++)
+    {
+      ctx.task(handle.read())->*[&](cudaStream_t stream, auto /*unused*/) {
+        cuda_sleep_kernel<<<1, 1, 0, stream>>>(clock_cnt);
+      };
+    }
+
+    ctx.task(handle.rw())->*[&](cudaGraph_t graph, auto /*unused*/) {
+      cudaGraphNode_t n;
+      cuda_safe_call(cudaGraphAddEmptyNode(&n, graph, nullptr, 0));
+    };
+  }
+
+  ctx.submit();
+
+  if (argc > 3)
+  {
+    std::cout << "Generating DOT output in " << argv[3] << '\n';
+    ctx.print_to_dot(argv[3]);
+  }
+
+  ctx.finalize();
+}