Improve InstantTranslate not causing excessive DB strain

Several problems with InstantTranslate:

• aborted requests in the Frontend may accumulate to "bunch of blocking requests "in the backend
• frontend should send new request only, when the request before it returned
• perspectively, frontend should request LRs independently (separate issue!)
• frontend should check, if text-content changed compared to last request
• investigation, if there is a bug in the frontend leading to parallel unaborted requests
• backend should check for request-abortion in the loop requesting LRs