Defect content drive search leaks info with limited user

[dario-daza]

Content Drive: Permission Filtering & Cursor-Based Pagination

Problem

Content Drive was not filtering results by user permissions, allowing users to see content they should not have access to. Additionally, the offset-based pagination strategy (startRow) skipped raw DB rows before applying permission filters, which produced sparse or empty pages — items were silently removed from an already-sliced result.

Solution

Content is now filtered by the requesting user's permissions before the page is assembled. The DB is scanned in configurable chunks; after each chunk, inaccessible items are discarded and only user-visible content is accumulated. Scanning stops as soon as enough visible items are collected to fill the page.

This guarantees that:

• Every page contains exactly maxResults items whenever enough visible content exists
• Users never see content they are not allowed to read
• Restricted users browsing large folders always get correctly sized pages, regardless of how many items are hidden from them

Cursor-based pagination
Both folders and content now use cursor-based pagination instead of an offset:

• contentCursor — the DB row position where the next content scan should resume
• folderCursor — the index into the folder list where the next folder page should resume

The request & response

Now returns nextContentCursor and nextFolderCursor, which clients pass on the next request to continue exactly where the previous page ended — no gaps, no duplicates.
Plus we also return hasMore and hasMore to know we could retrieve more content or folders for the next page.

Original Request	Proposed Request
Original Response	Proposed Response

This PR fixes: #34354

This PR fixes: #34354