EQUI-VOCAL: Synthesizing Queries for Compositional Video Events from Limited User Interactions - View it on GitHub