Add aiohttp to pyproject.toml dependencies

Version bump to 4.8.1 for aiohttp fix

Add ORCID integration to MCP server

- Add ORCID author search functionality
- Add ORCID publications retrieval
- Implement cross-validation capabilities
- Support name-based and affiliation-based search
- Version bump to 4.8.0

feat: Add PubMed search integration to MCP server

Added two new MCP tools for PubMed integration:
- search_pubmed: General PubMed search by author, DOI, title, or keywords
- pubmed_author_sample: Detailed author analysis with affiliations and variants

Features:
- Multiple search types (author, DOI, title, keywords)
- Institutional keyword extraction
- Name variant detection
- Email address extraction from affiliations
- Performance optimized with rate limiting and result caps
- Full XML parsing for detailed author information

Use cases:
- Cross-validation with OpenAlex data
- Author disambiguation support
- Institutional affiliation verification
- Name variant discovery for better matching

Version bump to 4.7.0

feat: Enhanced author disambiguation with filtering and institution-a…

…ware ranking

- Added intelligent filtering to remove candidates with no institutional affiliation
- Implemented institution-aware ranking using context matching
- Increased default autocomplete limit from 5 to 10 candidates
- Enhanced retrieve_author_works with complete retrieval options
- Added detailed logging for debugging and optimization
- Version bump to 4.6.0

Key improvements:
- Filter out 'No institution' candidates (reduces noise by ~40%)
- Context-aware ranking (improves accuracy for seasoned researchers)
- Complete works retrieval option for benchmark evaluation
- Institution matching with relevance scoring system

feat: Add autocomplete_authors tool for intelligent author disambigua…

…tion

- Add AutocompleteAuthorCandidate and AutocompleteAuthorsResponse models
- Implement autocomplete_authors_core using PyAlex autocomplete API
- Add autocomplete_authors MCP tool with context-based disambiguation
- Fast ~200ms response time with multiple candidate options
- Perfect for AI agents to select best match using institutional context
- Version bump to 4.5.0

Fix NoneType subscript error in filter_peer_reviewed_works

- Add None checking for work objects and titles
- Handle cases where work.get('title') returns None
- Add try-catch around individual work filtering
- Fixes 'NoneType' object is not subscriptable error

Add detailed error logging for NoneType debugging

Fix NoneType error in peer-review filtering

- Add safe None checking for title, journal_name, source_type, and work_type
- Convert None values to empty strings before calling .lower()
- Fixes 'NoneType' object is not subscriptable error for authors like Sankar Ghosh

Add configurable search_type parameter to search_works tool

- Add search_type parameter with options: 'general', 'title', 'title_and_abstract'
- Use title.search filter for precise title matching when search_type='title'
- Use title_and_abstract.search for title+abstract search
- Keep general search as default behavior for backward compatibility
- Update tool description and logging to include search_type info