I did see that, although it's plausible that an analogous change would be helpful in core postgres. The readme seemed to suggest that they tried this on cstore_fdw because that was easier on which to develop, not because there was more low-hanging fruit.