fix(tests): read changelog as UTF-8

[rioyu123]

Summary

Read CHANGELOG.md explicitly as UTF-8 in its test module.

The changelog now contains Unicode punctuation, but the tests used
Path.read_text() without an encoding. On Windows hosts whose locale encoding
is CP936/GBK, all five content-based changelog tests fail with
UnicodeDecodeError before reaching their assertions. UTF-8 is the repository
file encoding and is already used explicitly by other readers in the SDK.

This keeps the change limited to the five affected reads; test behavior is
otherwise unchanged.

Testing

• Before: 5 failed, 1542 passed, 12 skipped on Windows CP936
• After: 1547 passed, 12 skipped
• uv run ruff check tests/test_changelog.py
• uv run ruff format --check tests/test_changelog.py

Prepared with AI assistance and reviewed against the reported Windows failure;
the full test suite and focused lint/format checks were run locally.