1) Stat command does not work correctly if the plot group is
a model or a residual plot.


2) The sequence

fit 2
dg  2 res
r   y

should do a fit off.


3) XAX # does not work if # is a defined group.  (PLTXCG uses IOFSET
to compute starting location.  The correct PLT/isuba should be passed in.


4) Clean up the Xt(*) array, should always be a vector.


5) Users want a way to write out model subtracted data (WS?)


6) A (clever) user submitted bug1.qdp which should work.  Currently
xax # requires a vector not a group.  The documentation and some of
the code incorrectly said group, but for it to work correctly (not
crash) with the SKip command it needs to be a vector.  For it to use
a group, it must handle both the current SKip and the situations where
the X and Y groups don't have the same number of points.


7) Users should be allowed to control the delta chi^2 stopping condition.
(Currently hardwired in curfit.f).