v1.11.0

Added

• Bounded media evidence collection for local PDF text/metadata with page
  anchors, transcript cues with provenance, and image alt/caption context.
• Explicit optional OCR behavior and media-to-claim linkage validation through
  citable observe media --input <manifest> [--ocr].
• Medium-specific limitations that prevent extraction from being represented as
  claim support, reading-order validation, transcription accuracy, or semantic equivalence.