we should avoid the jank that happens on the first autocomplete search, when we try to clone the database connection.
no need to track this personally, should be fixed by unified complete btw.
Fwiw, this is fixed by UnifiedAutocomplete, I'd suggest to not modify the existing code and just wait for UnifiedComplete to be ready for prime time.
Removing this from the backlog again based on comment #2. We want this to be fixed by the UnifiedAutocomplete code instead of spending extra time on this alone.
Since UnifiedComplete is our future target of development, there's no reason to keep this open.