[Kst] update changes

[Brisset, Nicolas]

> > So.. you are basically suggesting that we revert the change 
> where we 
> > went to change in only one place...
> >
> > I'm not opposed to that - anyone else?
> 
>   Not opposed pending some solid regression testing.  The 
> legends were a problem for HFI until just recently so we need 
> to be sure they suffer no regressions.
I am aware that the legends were (and still are) a problem with respect
to printing and reloading files saved with a previous kst version, but
do you also mean there were usability issues linked with the legend
appearance attributes being present in the plot dialog ?