feat(linter): refactor parser to use remark for AST parsing

[davideast]

This pull request refactors the ParserHandler to use remark and the unified ecosystem instead of regular expressions for extracting YAML tokens from Markdown files. This makes the parser spec-compliant and robust against complex formatting and edge cases.

Key Changes

Core Refactor

• AST Parsing: Replaced string splitting and regex matching with full AST parsing using remark-parse and remark-frontmatter.
• Unified Pipeline: Implemented a unified processor to parse the document and traverse the tree with unist-util-visit.
• Robust Extraction: Now correctly handles YAML blocks with attributes (e.g., ```yaml title="theme") and handles indentation correctly as per Markdown specs.

Testing & Verification

• New Tests: Added a test for code blocks with attributes in handler.test.ts.
• Adversarial Testing: Added a test for long circular reference chains (10 steps) in model/handler.test.ts during a bug bash session, which passed successfully in 0.09ms.
• 100% Success: All 115 unit tests and the end-to-end fixture test pass successfully.

Dependencies Added

• unified
• remark-parse
• remark-frontmatter
• unist-util-visit
• @types/mdast (dev dependency for type resolution)