Bulk export via qTest search API returns duplicates and missing test cases across pages

Symptom

When iterating the qTest public REST API search endpoint (POST /projects/{id}/search with object_type filters) using page and page_size, the combined result set does not match the project contents: some test cases appear on two pages while others never appear at all, even though the total-count metadata stays consistent.

Context

The goal is a reliable full export of all test cases in a project for reconciliation against an external report. The search endpoint caps page_size (documented as 100 items per page in current guides), so the export must page through thousands of records. The loop runs while other team members continue editing and creating test cases in the UI, and each edit also creates a new test case version.

My working assumption is that the paginated search is not snapshot-isolated, so items created or re-sorted mid-iteration shift between pages. I have not found documentation stating whether a stable cursor, sort tiebreaker, or point-in-time token exists for the search API. Assume a current SaaS qTest Manager tier with /api/v3 available.

Questions

1. Is non-snapshot pagination the documented behavior of the qTest search API, or does a cursor/snapshot mode exist that guarantees a consistent result set?
2. Does specifying a stable sort attribute (e.g., creation date with an ID tiebreaker) eliminate duplicates and omissions during concurrent edits?
3. Is there a supported bulk-export alternative to paginated search for large projects?