Measurement sets: FLAG_ROW column of ANTENNA table is unutilised

[cjordan]

At present, there is no unambiguous way to know if a main table row (which contains data for a baseline) is completely flagged because one of its tiles is flagged, or if the timestep is flagged. Therefore, in hyperdrive, I must consider all tiles as unflagged, even though Birli correctly flags baselines containing a flagged tile.

One hopefully neat way to improve this situation is to use the FLAG_ROW column of the ANTENNA table. Perhaps to match cotter's behaviour, this is always false, but I don't see the harm in allowing these values to be true. If this happens, then hyperdrive is able to unambiguously detect flagged tiles and use less memory for its operations with these MSs.