Two views of the trending corpus: cells sized by star count and by momentum score, grouped and colored by language. Drawn from the same pipeline that feeds /signals and /consensus.