feat(bench): add tracemalloc memory benchmarks for parser modules

[Oaklight]

Summary

• Adds test_<module>_memory.py for soup, yaml, xml, multipart, readability, and protobuf (closes Benchmark: add memory peak measurement for parser modules #84, closes Benchmark: add memory peak measurement for parser modules #85)
• Uses stdlib tracemalloc only — no new dependencies, no pytest-benchmark
• Each file covers S/M/L input sizes consistent with the existing time benchmarks
• Reports peak allocation in KB; zerodep-vs-reference comparison tests are included wherever the reference library is available (bs4, PyYAML, xmltodict, readability-lxml); skipped otherwise
• protobuf: google-protobuf comparison tests exist but are skipped when the MessageFactory.GetPrototype API is unavailable in the installed version

Test plan

• conda run -n zerodep pytest soup/test_soup_memory.py yaml/test_yaml_memory.py xml/test_xml_memory.py multipart/test_multipart_memory.py readability/test_readability_memory.py protobuf/test_protobuf_memory.py -v -s → 74 passed, 2 skipped
• ruff check --fix → clean
• ruff format → no changes
• pre-commit hooks (ruff, ty, complexipy) → all passed