Add PDF scrapers with URL resolution

[muthukumaranR]

Summary

Add PDF scraper capabilities with URL resolution and fallback strategies.

Changes

• PyPaperBot Integration: Multi-source PDF retrieval with CrossRef/Scholar/Unpaywall fallbacks
• URL Resolution: ArXiv abs → pdf conversion, publisher-specific resolvers
• Waterfall Strategy: Sequential scraper execution with fallbacks
• Unpaywall API: DOI-to-open-access resolution with email configuration

Usage

# UnpaywallResolver for DOI resolution
config = ArticleResolverConfig(debug=True, user_agent="Research Bot")
resolver = UnpaywallResolver(config=config)
result = await resolver.arun(ResolverInputSchema(url="https://doi.org/10.1038/nature12345"))

# PyPaperBot scraper
scraper = PyPaperBotScraper()
result = await scraper.arun(ScraperToolInputSchema(url="https://arxiv.org/abs/2301.00001"))

# Waterfall scraper with URL resolution
waterfall = WaterfallScraper()
result = await waterfall.arun(ScraperToolInputSchema(url="https://doi.org/10.1038/example"))

# CompositeWaterfallScraper with PyPaperBot fallback
composite = CompositeWaterfallScraper(scrapers=["docling", "pypaperbot"])
result = await composite.arun(ScraperToolInputSchema(url="https://arxiv.org/pdf/2301.00001.pdf"))

Testing

• Integration tests with real API calls for UnpaywallResolver
• PyPaperBot configuration and DOI extraction tests
• WaterfallScraper URL resolution logic validation
• CompositeWaterfallScraper fallback behavior tests

[NISH1001]

Can we move the composite scrapers like this to akd.tools.scrapers.composite because it also uses waterfall as these are composite scrapers with dependency injection.

[NISH1001]

Isn't matolotlib extra dependency?