feat(cccl): integrate missing CCCL directories — python/, ci/, .agent/, docs/, test/

Sparse-checkout from NVIDIA/cccl main branch to complete cccl_upstream:

Added:
- python/cuda_cccl/ (226 files) — Python bindings for device-level algorithms
  Critical for muh toolchain: cuda.compute.reduce_into, scan, radix_sort, etc.
  Includes 204 .py files with full test coverage for all 27 algorithms
- ci/ (163 files) — Build/test infrastructure
  build_cub.sh, test_cub.sh, build_and_test_targets.sh, matrix.yaml
  Directly maps to our [INFRA-CI] and [INFRA-BUILD] items
- .agent/skills/ (7 files) — NVIDIA's own agent skills for CCCL
  cccl-style/SKILL.md, cccl-test/SKILL.md, sass-diff/SKILL.md
- docs/ (491 files) — Official CCCL documentation
  CI references, CMake guides, Python compute docs, libcudacxx PTX docs
- test/ (12 files) — Top-level integration tests (cuda_smoke, stdpar)
- Root configs: .clang-format, .clang-tidy, CONTRIBUTING.md, pyproject.toml
- CLAUDE.md symlink → AGENTS.md (NVIDIA's standard)

cccl_upstream now mirrors full NVIDIA/cccl structure:
  Before: 42M (cub + thrust + libcudacxx + cudax + c + examples + benchmarks)
  After:  53M (+python +ci +docs +.agent +test +configs)

This completes the CCCL base needed for:
- [muh-bench] items: ci/util/build_and_test_targets.sh for targeted builds
- [CCCL-verify] items: python/cuda_cccl/tests/ as reference implementations
- [CCCL-test] items: ci/test_cub.sh, ci/test_thrust.sh
- Agent workflow: .agent/skills/ for consistent style and test patterns

cccl_upstream/.pre-commit-config.yaml

@@ -0,0 +1,143 @@
+# Copyright (c) 2024, NVIDIA CORPORATION.
+ci:
+    autofix_commit_msg: |
+      [pre-commit.ci] auto code formatting
+    autofix_prs: false
+    autoupdate_branch: ''
+    autoupdate_commit_msg: '[pre-commit.ci] pre-commit autoupdate'
+    autoupdate_schedule: quarterly
+    skip: [mypy, secret-scan-trufflehog] # mypy fails on pre-commit.ci due to numpy stub incompatibility with Python 3.14
+    submodules: false
+
+repos:
+  # Runs first so a leaked credential blocks the commit before any formatter runs.
+  # Self-installing: the hook downloads a pinned, checksum-verified trufflehog on
+  # first use (no manual install). Skipped on pre-commit.ci; Pulse CI enforces server-side.
+  - repo: https://github.com/NVIDIA/security-workflows
+    rev: v0.2.0
+    hooks:
+      - id: secret-scan-trufflehog
+
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v5.0.0
+    hooks:
+      - id: end-of-file-fixer
+      - id: mixed-line-ending
+      - id: trailing-whitespace
+        # handled by clang-format
+        exclude_types: [c, c++, cuda]
+      - id: check-json
+      - id: check-toml
+      - id: pretty-format-json
+        args: ['--autofix', '--indent=2', '--no-sort-keys']
+      - id: check-symlinks
+      - id: check-executables-have-shebangs
+      - id: check-merge-conflict
+      - id: check-shebang-scripts-are-executable
+        # ruff checks this already
+        exclude_types: [python]
+      - id: check-yaml
+
+  - repo: https://github.com/shellcheck-py/shellcheck-py
+    rev: v0.11.0.1
+    hooks:
+      - id: shellcheck
+        exclude: |
+          (?x)^(
+            ^.*libcudacxx/cmake/config\.guess$
+          )
+
+  - repo: https://github.com/pre-commit/mirrors-clang-format
+    rev: v20.1.7
+    hooks:
+      - id: clang-format
+        types_or: [file]
+        files: |
+          (?x)^(
+            ^.*\.c$|
+            ^.*\.cpp$|
+            ^.*\.cu$|
+            ^.*\.cuh$|
+            ^.*\.cxx$|
+            ^.*\.h$|
+            ^.*\.hpp$|
+            ^.*\.inl$|
+            ^.*\.mm$|
+            ^libcudacxx/include/.*/[^.]*$
+          )
+        args: ["-fallback-style=none", "-style=file", "-i"]
+
+  # TODO/REMINDER: add the Ruff vscode extension to the devcontainers
+  # Ruff, the Python auto-correcting linter/formatter written in Rust
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.12.2
+    hooks:
+    - id: ruff  # linter
+    - id: ruff-format  # formatter
+
+  # CMake formatting
+  - repo: https://github.com/BlankSpruce/gersemi
+    rev: 0.23.1
+    hooks:
+    - id: gersemi
+
+  - repo: https://github.com/codespell-project/codespell
+    rev: v2.4.1
+    hooks:
+      - id: codespell
+        additional_dependencies: [tomli]
+        args: ["--toml", "pyproject.toml", "-I", ".codespell-ignore.txt"]
+        exclude: |
+          (?x)^(
+            build|
+            CITATION.md
+          )
+
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: 'v1.16.1'
+    hooks:
+      - id: mypy
+        # numpy is intentionally not installed here: numpy>=2.3 ships PEP 695
+        # `type` statements in its stubs that mypy rejects under
+        # python_version=3.10, and numpy<2.3 has no wheels for the Python that
+        # pre-commit.ci runs. numpy is instead ignored in the mypy config.
+        additional_dependencies: [types-cachetools]
+        args: ["--config-file=python/cuda_cccl/pyproject.toml",
+               "python/cuda_cccl/cuda/compute/"]
+        pass_filenames: false
+
+  - repo: https://github.com/sirosen/texthooks
+    rev: 0.7.1
+    hooks:
+      - id: fix-smartquotes
+      - id: fix-spaces
+      - id: forbid-bidi-controls
+      - id: fix-ligatures
+
+  - repo: local
+    hooks:
+      - id: check-shebang
+        name: check-shebang
+        entry: ci/util/pre-commit/check_shebang.py
+        language: python
+        types: [shell]
+        exclude: |
+          (?x)^(
+            ^.*libcudacxx/cmake/config\.guess$
+          )
+
+      - id: check-cub-test-macros
+        name: require CUB test memory classification
+        entry: ci/util/pre-commit/check_cub_test_macros.py
+        language: python
+        files: ^(cub/test/.*\.(cu|cuh|h)|ci/util/pre-commit/check_cub_test_macros\.py)$
+        types: [file]
+
+      - id: unprintable-unicode
+        name: unprintable-unicode
+        entry: ci/util/pre-commit/strip_unprintable.py
+        language: python
+        types: [text]
+
+default_language_version:
+  python: python3