Skipped binary inline-image payload bytes after the PDF ID operator so tokenizer scanning resumes at post-image content.
Advanced past stray closing delimiters while recovering malformed content streams to avoid zero-width tokenizer iterations.
Fixes#4512
