Documentation on early

Task: early
Purpose: Flag data which has early/normal problems.
Categories: calibration, uv analysis

        EARLY is a Miriad task which flags visibility data that it suspects
        contains early/normal problems. The input must be a multi-channel
        dataset (NOT channel 0). Early works by comparing the scatter of
        data within a visibility record with that expected because of thermal
        noise. If the scatter exceeds some number of sigmas, then the data
        are assumed to be affected by the early/normal problem, and the
        entire record is flagged.

        Several assumptions about the data are make:
        Firsly there is significant signal in the visibility
        record so that the phase slope across the band causes a substantial
        increase in the scatter. This will not be true if the signal is
        weak, and the visibility records are noise dominated. If the signal
        is noise dominated, no early/normal problems can be detected, and
        the data will not be flagged.

        Secondly the data have been bandpass calibrated. EARLY always applies
        a bandpass correction to the data if there is one present.

        Thirdly that the noise level is fairly constant across the band,

        Fourthly that any end channels that might be bad have already been
        flagged.

        Fifthly there is no substantial variation of the signal across
        the band. This will be true for continuum, but not spectral
        line work.

Key: vis
        The input visibility data-set to flag. Several files can be
        given. Wildcards are supported. No default.

Key: select
        Normal uv selection. This selects which data to check for early/normal
        problems.

Key: sigma
        Number of sigmas. Default is 10.

Key: log
        Output log file for messages. The default is the terminal.

Key: options
        Extra processing options. Several can be given, separated by commas.
        Possible values are:
          noapply    Do not do any flagging -- just check the data, and
                     produce the appropriate messages.
          verbose    Give details about what is (or would be) flagged.