MI-T4: Markdown import (.md → new page, hydrate-on-open)

[coderhd]

Parent

#26 — H0 — Markdown Import/Export (full round-trip)

What to build

A .md file becomes a new page: frontmatter → title + pages.properties + tags; body → rich editor content written through the existing collaboration path (hydrate-on-open — no new server route). Imported [[wikilinks]]/#tags are indexed by the graph index service like normal edits, and the server-side graph index folds properties.tags into the tag index so imported frontmatter tags are searchable.

Acceptance criteria

• A dashboard Import action accepts a .md file; import creates a page and navigates the user to it.
• services/import.ts parses the file: title from frontmatter (fallback: filename), frontmatter keys → pages.properties, tags → properties.tags; parses the body via the round-trip engine; creates the page; stashes the body in a per-tab payload store keyed by page id.
• EditorWorkspace accepts initialContent and hydrates the editor on first mount only when the doc is empty; the payload is cleared after consumption; a refresh before hydration just opens the page empty.
• server/graph-index.js indexPage reads pages.properties and folds properties.tags (array or comma/space-separated string) into the tag rows written via sync_page_graph; body tags and wikilinks still extracted; links resolution unchanged. sync_page_graph remains service-key-only.
• Success toast on import; clear error (no page created) for empty/malformed files; frontmatter-less files import cleanly.
• Viewer/read-only surfaces show no import action.
• Focused unit tests pass (import service, hydration helper, graph-index tag fold); npm run lint and npm run build green.