feat: expose subscript and superscript markers in CLI and MCP

[Çağdaş Yürekli (cagdasyurekli)]

Fixes #1217.

The Python API already forwards markdownify's sub_symbol and sup_symbol options for DOCX, but the CLI and MCP tool do not expose them. Add --sub-symbol / --sup-symbol to the CLI and optional string arguments with the same names to convert_to_markdown, and document all three interfaces.

For example, a DOCX containing subscript/superscript runs can now produce H<sub>2</sub>O and x<sup>2</sup> through the CLI or MCP instead of H2O and x2. Both file and stdin CLI input are supported. Omitting the options preserves current behavior; custom markers and explicit empty strings are supported. This closes the remaining interface/discoverability gap described in the issue discussion without adding another converter or changing PDF/cloud extraction.

Validation on Python 3.12:

• GITHUB_ACTIONS=true hatch test --python 3.12 -q --tb=short: 871 passed, 43 skipped (the repository's conditional skips).
• CLI suite: 11 passed, including real minimal DOCX conversions from both files and stdin with default, HTML, custom, and empty markers.
• Full MCP suite: 22 passed, including real stdio tool discovery and calls, optional schema fields, null/default arguments, existing protocol compatibility, HTTP transports, and error handling.
• pre-commit run --all-files and git diff --check: passed.