bin_size, error_power, do_radial
Set the type and degree of weighting for subsequent UV gridding.
EXAMPLE
-------

1. To select uniform weighting and to scale the uniform weighting grid
   weights by the errors implied by the data raised to the power -1, type:

   0>uvweight 2,-1
   Uniform weighting binwidth: 2 (pixels).
   Gridding weights will be scaled by errors raised to the power -1.
   Radial weighting is not currently selected.
   0>

   The first line illustrates a typical use of the command, and the
   following three lines are the response by the 'uvweight' function,
   confirming the choices made.

2. To see the current choice of weighting, omit all arguments:

   0>uvweight
   Uniform weighting binwidth: 2 (pixels).
   Gridding weights will be scaled by errors raised to the power -1.
   Radial weighting is not currently selected.
   0>

PARAMETERS
----------
bin_size  - If this is > 0 then it selects uniform weighting. The
            value sets the bin size in UV-grid pixels within which to
            count neighboring visibilities. The reciprocal of these
            counts are then used to weight the visibilities by the
            local number density. A value of 2 is recommended.
            Non-integral values are legal but fractional values below
            1 are not.

error_pow - If < 0 then this selects additional weighting by amplitude
            errors. The value selects the exponent to raise the error
            to. Statistically this should be -2, but -1 is recommended.
            Make sure that the relative errors between baselines are
            meaningful before using this option or you may end up with
            some baselines making no contribution to the final map!

do_radial - If true then the weight given to each visibility will be
            divided by its UV radius. Only recommended when one's
            UV coverage is approximately circular.

CONTEXT
-------
This command sets the weights given to individual visibilities as they
are gridded, prior to Fourier inversion into the image plane. The
choices made remain in effect until this function is called again.

When a new data set is read with the 'observe' command, the defaults
are reset to apply uniform weighting with a bin-width of 2 and no
other form of weighting.

Unless the commands mentioned in the last section are of interest, you
would typically follow this command by invoking the 'invert' command.
(See help on 'invert' for details).

SIDE EFFECTS
------------
Use of this command will cause the next 'clean', 'mapplot' etc..
to re-invert the modified UV data-set to produce an updated dirty
beam and residual map.

RELATED COMMANDS
----------------
uvtaper - Apply a gaussian taper to the weights.
uvrange - Constrain the min and max UV radii within which visibilities
          are gridded.
uvzero  - Set a zero-UV-spacing flux.