Add Python 3.13 support

[JoelNiklaus]

Problem

Datatrove only declares and tests support for Python 3.10, 3.11, and 3.12, even though Python 3.13 (Oct 2024) is in active maintenance. Users on newer Python versions see missing classifiers on PyPI and have no CI coverage guaranteeing compatibility.

Solution

• Add Programming Language :: Python :: 3.13 classifier to pyproject.toml.
• Extend the CI test matrix in .github/workflows/testing.yml to include 3.13.
• Modernize type annotations across 11 files: replace deprecated typing.Optional, typing.List, typing.Dict, typing.Tuple, and typing.Union with built-in generics (X | None, list, dict, tuple, X | Y) and collections.abc (Callable, Sequence). The project already targets >=3.10, so these are all safe.
• Add "Requires Python 3.10+." note to the README installation section.

Notes

Python 3.14 is not included yet because key optional dependencies (tensorflow, vllm) don't ship cp314 wheels. Since requires-python = ">=3.10.0" already permits 3.14 installs, users can still install and use datatrove on 3.14 -- we just don't test it in CI yet.

Testing

• All 347 tests pass on Python 3.12 (current venv).
• Ruff linting and formatting checks pass across all modified files.
• Full CI validation on 3.13 will run when this PR is opened.