Expose byte metadata from HtmlHandlers.extract_html_content #6
Description
Summary
Layer 3's binary optimization still has to manually convert the grapheme-based chars_consumed coming back from HtmlHandlers.extract_html_content/2 into byte offsets. We just fixed one caller by recomputing the byte span, but each caller has to copy that logic.
Proposed Fix
Have HtmlHandlers.extract_html_content/2 return both grapheme and byte counts (for example {content, chars_consumed, bytes_consumed}) so every caller can rely on a single source of truth. Update the IO-list and binary paths to use the new field and drop their ad-hoc math.
Benefits
- Eliminates duplicated slicing logic
- Prevents future drift when we add new consumers
- Makes it harder to regress on multi-byte graphemes (CRLF, emoji, etc.)