Improve forwarded newsletter extraction and reading

[kirahowe]

Summary

• extract original newsletter identity, authorship, metadata, and content from common forwarded-email formats
• strip forwarding chrome while retaining conservative fallbacks for ordinary replies
• persist extraction provenance and add safe, versioned reprocessing from retained raw email
• harden newsletter rendering for narrow mobile viewports and oversized email markup
• add privacy protections for remote newsletter content
• provide a bounded production reprocessing command and runbook

Verification

• 239 tests, 1,425 assertions, 0 failures
• clj-kondo clean
• cljfmt clean
• production uberjar built with the reprocessing entrypoint
• browser checks at 390px and 320px, including oversized tables and long unbroken content