feat(web-fetch): extract text from PDF documents

[ZhiXiao-Lin]

Summary

• detect PDF responses by media type or file signature and extract their text in web_fetch
• run PDF parsing off Tokio worker threads and return clear errors for malformed or image-only documents
• expose normalized content type and document kind in tool metadata while preserving HTML redirect handling
• add in-memory PDF coverage for detection, extraction, malformed files, and empty documents

Validation

• cargo fmt --all -- --check
• cargo clippy --workspace --lib --bins -- -D warnings
• cargo test --workspace
• cargo test --workspace --all-features --lib
• git diff --check