UVFLAG can flag using autocorrelation data

[VincentMcIntyre]

Add selection and baseline removal options to uvflag to facilitate flagging RFI based on autocorrelation data. Some amount of experimentation with real data will be required to determine the exact approach to take.

See Changes/uvflag

[MarkWieringa]

Two parts:

1. Select and identify RFI in auto corr spectra using uvflag
   • May need to take bandpass into account
2. Write uvacflag to transfer flags from autocorr data to crosscorr data.

[MarkWieringa]

Part 2 is complete - autocorr flags can now be transferred to cross correlations using task uvacflag

First impression from some std correlator data with autocorrelations is that this may not be as useful as it sounds. The dynamic range on rfi is much better in the cross corr data, than the auto corr data. So unless we can remove the large 'zero-spacing' baseline using a low order fit it may not help much. More research required using real CABB data to make further progress on this.

[MarkWieringa]

Turns out the CABB bandpass has a lot of structure, making identification of all but the strongest RFI in the autocorrelations pretty hard. We're probably better off with a combination of online CABB flags for fixed channel problems and the atlod rfiflag option for fixed frequency RFI. The rest is hard labour with a combination of uvflag, blflag, pieflag or tvclip.