fix(parsers): improve Python requirements verification coverage

[mstykow]

Summary

• broaden pip requirements discovery so nested requirements/** and requirements-txt/** inputs are recognized, and regenerate docs/SUPPORTED_FORMATS.md
• link nested Python requirements files to the nearest project package during assembly and add scan-level regression coverage for nested and standalone requirements trees
• record the verified astral-sh/uv benchmark result and clarify that benchmark rows hold end-state advantages while scorecard verification normally changes only the Status cell

Scope and exclusions

• Included:
  • Python requirements parser path matching
  • Python requirements assembly assignment and scan tests
  • benchmark and scorecard authoring rules for verification updates
• Explicit exclusions:
  • no new parser golden fixtures or output-format fixtures
  • no broader Python/PyPI verification targets beyond the recorded astral-sh/uv reference run

Intentional differences from Python

• None for parser behavior; the code changes are parity-oriented and the documentation changes only clarify how verified outcomes are recorded.

Follow-up work

• Created or intentionally deferred:
  • continue the remaining Python / PyPI verification targets in the scorecard using the clarified benchmark-versus-scorecard workflow

Expected-output fixture changes

• Files changed: None.
• Why the new expected output is correct:
  • no expected-output fixtures changed; behavior was validated with targeted parser and scan tests plus the saved astral-sh/uv compare run