test: add prompt injection regression fixtures

[AKilalours]

Summary

This PR adds prompt injection regression fixtures for AI-assisted security review workflows.

Why

The repository already notes that the action is not hardened against prompt injection attacks, and PR #105 adds a threat model for that risk. This follow-up adds concrete adversarial fixtures that can be used to reason about instruction override attempts in Markdown, code comments, workflow configuration, and false-positive filtering contexts.

Changes

• Added docs/prompt_injection_regression_fixtures.md
• Added Markdown instruction override fixture
• Added code comment suppression fixture
• Added workflow configuration tampering fixture
• Added false-positive filter suppression fixture

Testing

• Documentation and fixture-only change
• Checked Markdown formatting locally
• Commit signed and verified

[AKilalours]

Hi maintainers,

I opened this as a follow-up to the prompt injection threat model PR. This adds concrete adversarial fixtures for Markdown, code comments, workflow configuration, and false-positive filtering contexts.

Happy to revise the file location, naming, or scope if you prefer a different structure.