refactor: replace dict-based query with ParsedQuery dataclass

[filipchristiansen]

This PR replaces the old dictionary-based query object with a new ParsedQuery dataclass, improving type safety and readability across the codebase.

Key points:

• Introduces a ParsedQuery dataclass that consolidates all query-related fields (repo, branch, commit, patterns, etc.)

• Updates the following functions in the query_parser module to return a query of type ParsedQuery (instead of dict[str, Any]):

  • _parse_repo_source
  • _parse_path
  • parse_query

• Updates the following functions in the query_ingestion module to take a query argument of type ParsedQuery (instead of dict[str, Any]):

  • run_ingest_query
  • _ingest_directory
  • _ingest_single_file
  • _create_tree_structure
  • _create_summary_string
  • _extract_files_content
  • _process_item
  • _process_symlink
  • _scan_directory

• Switches ignore/include patterns to sets for clearer overrides and deduplication

• Moves or imports maximum size constants (MAX_FILE_SIZE, etc.) to config.py

• Aligns all references and tests to the new dataclass approach

This unification makes it easier to understand and maintain how query data flows through the ingestion process.

[cyclotruc]

Tested it, looks good