Currently we are only using a single reference antenna to initialise delays. This can lead to the delay estimate failing when there are severely flagged baselines to the given reference antenna. As discussed it should be possible to use multiple antennas to improve the delay estimates. I suggest iterating through the least flagged antennas until all antennas have valid estimates (which may include zero, valid here means that there is a peak in the power spectrum). This has the added benefit that it saves the user from specifying a reference antenna
Currently we are only using a single reference antenna to initialise delays. This can lead to the delay estimate failing when there are severely flagged baselines to the given reference antenna. As discussed it should be possible to use multiple antennas to improve the delay estimates. I suggest iterating through the least flagged antennas until all antennas have valid estimates (which may include zero, valid here means that there is a peak in the power spectrum). This has the added benefit that it saves the user from specifying a reference antenna