Normalize punycode labels in extraction results

[gaoflow]

Summary:

• Decode valid xn-- labels before assembling extraction result fields so Unicode and ACE forms produce the same suffix/fqdn.
• Preserve invalid punycode labels unchanged.
• Add regression coverage for equivalent Unicode and punycode IDN suffixes.

Tests:

• PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 .tox/py310/bin/python -m pytest tests/main_test.py -k puny -q
• .tox/py310/bin/ruff format .
• uv run tox -e py310 -- -p no:gitignore
• uv run tox -e lint
• uv run tox -e typecheck
• uv run tox -e codestyle

Note: the local fresh tox environment installs pytest 9.1.1 and pytest-gitignore currently fails during plugin loading, so the py310 run disables only that plugin with -p no:gitignore.

Fixes #366

AI assistance was used under my direction.