rockskip: Drop DISTINCT from query

Created by: chrismwendt

This DISTINCT was unnecessary because we'll get one symbol per row anyway, and the limit is exactly what we want. I think this was a vestige of the blob-oriented data structure.

It was causing searches to be really slow (22s on the megarepo). Afterwards, they're ~120ms.

Test plan

Ran it and made sure it returns multiple symbols per path.