Add veraPDF corpus and progressive testing strategy wiki pages
Document Challenge 7: Sequential Redaction and Bounding Box Accumulation Added comprehensive documentation of the root cause for issue #124: - Sequential redactions cause massively inflated bounding boxes - TextSegmenter incorrectly calculates width for unmapped characters - Reconstructed operations lose spatial information - Width accumulation leads to 1000+ point bboxes (page is only 612!) - Inflated bboxes intersect with unrelated redaction areas - 95% of content incorrectly removed in real-world tests Includes: - Detailed root cause analysis with code locations - Evidence from debugging logs - Recommended fix (use fixed width for unmapped chars) - Lessons learned about reconstructed operations - Verification test pattern for sequential redactions - References to related issues #124 and #125
docs: Initialize GitHub wiki with PDF documentation Wiki pages covering: - PDF Text Operators (Tj, TJ, ', ") - PDF Coordinate Systems - Font Types and Encodings - Content Stream Structure - Project Architecture - Redaction Engine - Testing Strategies - Lessons Learned (scrambled content, coordinate conversion) 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>