fix(core): Add `.geminiignore` support to SearchText tool

[xyrolle]

The SearchText (ripgrep) tool was not respecting .geminiignore patterns, causing subagents to search through files that users explicitly wanted to exclude. This fix adds .geminiignore support by leveraging ripgrep's native --ignore-file flag, ensuring consistent ignore behavior across all tools

Details

• Extended GeminiIgnoreParser with getIgnoreFilePath() to return the .geminiignore path when it exists and has valid patterns, and hasPatterns() to check if any patterns are defined
• Added an existence check in getIgnoreFilePath() to handle the race condition where the file could be deleted after the parser is instantiated
• Updated RipGrepTool to cache the GeminiIgnoreParser instance, avoiding redundant file I/O on repeated tool calls
• Ripgrep natively handles .gitignore, so no changes needed there
• Added 8 unit tests for the new 1GeminiIgnoreParser` methods

Related Issues

Partially Fixes: https://github.com/google-gemini/gemini-cli/issues/13651

How to Validate

1. Create a file (example file.md) with 1 word "carrot" in it
2. add this file to .geminiignore
3. gemini "please find "carrot" in the project"
4. observe that result is not found

behaviour before fix:

• I found one instance of 'carrot' in file.md. file.md*: Line 1 (carrot)

Pre-Merge Checklist

• [ ] Updated relevant documentation and README (if needed)
• [x] Added/updated tests (if needed)
• [x] Noted breaking changes (if any)
• [ ] Validated on required platforms/methods:
  • [x] MacOS
    • [x] npm run
    • [x] npx
    • [ ] Docker
    • [ ] Podman
    • [ ] Seatbelt
  • [ ] Windows
    • [ ] npm run
    • [ ] npx
    • [ ] Docker
  • [ ] Linux
    • [ ] npm run
    • [ ] npx
    • [ ] Docker