feat(ingest): improve book reading quality and import hardening

[jdsalasca]

Summary

• harden book import pipeline: extension whitelist (.txt/.pdf), readability checks and max-size validation
• add configurable import size limit (�pp.import.max-bytes)
• improve PDF extraction consistency: per-page extraction with page separators and position sorting
• improve normalizer quality: remove TOC/index noise and merge wrapped paragraph lines
• add unit tests covering new import/normalization/extraction behavior

Validation

• mvn test