feat: chat uploads, vision turn isolation, and MindSpace agent improvements

Add chat file/image upload UX, attachment proxying, vision thumbnails, and per-turn image scoping so agents only use the current upload. Extend MindSpace asset context, billing token state, OA/scenario verify scripts, and related runtime config.

Co-authored-by: Cursor <cursoragent@cursor.com>

mindspace-assets.mjs

@@ -264,6 +264,7 @@ function publicTempImageUrl(userId, assetId, mimeType, fallbackFilename = '') {
 }
 
 function assetResponse(row) {
+  const workspaceRelativePath = normalizeWorkspaceRelativePath(row?.workspace_relative_path);
   return {
     id: row.id,
     categoryId: row.category_id,
@@ -282,6 +283,7 @@ function assetResponse(row) {
     hasThumbnail: Boolean(row.has_thumbnail ?? row.mime_type === 'text/html'),
     publicUrl: publicUrlForAsset(row),
     sourcePageId: row.source_page_id ?? null,
+    ...(workspaceRelativePath ? { workspaceRelativePath } : {}),
     createdAt: asNumber(row.created_at),
     updatedAt: asNumber(row.updated_at),
   };