What should to_parity's reduce_col_deg default be?
It's not obvious to me why reduce_col_deg
should be disabled by default. Do we have statistics to backup this choice?
From the CAR point of view, if the degeneralization increase the number of states without decreasing the number of colors, it is unlikely to help producing fewer states after CAR.
From the IAR point of view, it could be the opposite: the patterns that partial degeneralizations remove cannot be part of Rabin-like or Streett-like formulas (except very small corner cases), so applying partial degeneralization even if the number of colors does not change might help using IAR instead of CAR.