Summary

Fixes #2305

Image wikilinks like ![[my_photo_2024-01-01_overview_(1).jpg]] fail to render as <img> tags because the Markdown parser interprets _text_ patterns in the filename as emphasis, fragmenting the wikilink text across multiple mdast nodes before mdastFindReplace can match and convert it.

This PR modifies the existing textTransform wikilink handler in ofm.ts to convert image embeds to standard Markdown image syntax with angle-bracket destinations (![alt](<filename>)), which preserves underscores verbatim through the Markdown parser.

Root cause

The processing pipeline in ofm.ts:

1. textTransform — recognizes ![[file.jpg]] via wikilinkRegex, but re-emits it as raw ![[file.jpg]] text
2. remark parser — parses the text into mdast; _text_ patterns inside [[...]] become emphasis nodes, splitting the wikilink text
3. markdownPlugins / mdastFindReplace — can no longer match the wikilink regex because the text is fragmented by emphasis nodes

Fix

In the textTransform phase where image wikilinks are already identified by wikilinkRegex, convert them to standard Markdown image syntax:

![[file_name_(1).jpg]]       →  ![](<file_name_(1).jpg>)
![[file_name_(1).jpg|alt]]   →  ![alt](<file_name_(1).jpg>)

The angle-bracket destination syntax is part of the CommonMark spec and prevents the Markdown parser from interpreting underscores as emphasis. CrawlLinks resolves the image path from the resulting image node as usual.

Non-image embeds (video, audio, PDF, transclusion) remain as wikilinks for the existing markdownPlugins handling.

Test plan

• Image wikilinks with underscores render correctly (e.g. ![[my_photo_2024.png]])
• Image wikilinks without underscores still render correctly
• Image wikilinks with aliases render correctly (e.g. ![[photo.png|alt text]])
• Non-image embeds (video, PDF, transclusion) are unaffected
• Images inside callouts render correctly