feat: add legacy .doc conversion support

[mikemikimike]

Related issue

Fixes #23

Summary

Adds optional conversion support for legacy binary .doc files using the unword parser.

Implementation

• Added DocConverter, which recognizes .doc files and returns the parser's body text as Markdown.
• Added the doc optional dependency and included it in the all feature.
• Registered the converter independently from the existing .docx converter.
• Added tests for extension detection, successful conversion, and missing dependency handling.

Compatibility

Existing .docx and other converters are unchanged. .doc support is opt-in through markitdown[doc] or included in markitdown[all]. Rich layout extraction is limited to the text exposed by unword.

Validation

• python -m pytest tests/test_doc_converter.py -q — 3 passed.
• python -m mypy --ignore-missing-imports src/markitdown/converters/_doc_converter.py tests/test_doc_converter.py — passed.
• python -m black --check src/markitdown/converters/_doc_converter.py tests/test_doc_converter.py — passed.
• python -m pre_commit run --all-files — passed.
• python -m pytest -q --basetemp=.pytest-tmp — not fully green in this Windows environment: existing CLI vector tests encounter GBK decoding errors, and test_file_uris expects Unix-style paths. The focused and unaffected tests passed before those environment-specific failures.

Limitations

No binary .doc fixture is included; extensionless MIME sniffing and rich layout reconstruction are out of scope.