feat: Add split table merging support to EnhancedPDFParser

[AdemBoukhris457]

Summary

This PR adds split table merging functionality to EnhancedPDFParser, enabling automatic detection and merging of tables that span across multiple pages. This feature was previously only available in StructuredPDFParser.

Changes

Code Changes

• doctra/parsers/enhanced_pdf_parser.py:
  • Added merge_split_tables parameter and related configuration options to __init__
  • Implemented split table detection logic in _process_parsing_logic
  • Added logic to skip individual table segments that are part of merged tables
  • Implemented processing and saving of merged table images
  • Added notification message when split tables are detected

Documentation Changes

• docs/user-guide/parsers/enhanced-parser.md:

  • Added split table merging to key features
  • Added comprehensive section documenting the feature
  • Included configuration examples and parameter details
  • Added guidance on when to use the feature

• docs/api/parsers.md:

  • Updated EnhancedPDFParser quick reference with split table merging parameters
  • Added note that feature is available for both parsers

Features

• Automatic detection of tables split across page boundaries using two-phase approach:
  • Phase 1: Proximity detection (spatial heuristics)
  • Phase 2: Structural validation (LSD-based column alignment)
• Merged tables are saved as single composite images
• Full VLM support for merged table extraction
• Configurable thresholds and confidence scores
• Comprehensive error handling

Configuration

The feature can be enabled by passing merge_split_tables=True when creating an EnhancedPDFParser instance:

parser = EnhancedPDFParser(
    use_image_restoration=True,
    merge_split_tables=True
)