Slowing scatter down for the sake of Matlab compatibility

[John W. Eaton]

On  4-Dec-2011, Jordi Gutiérrez Hermoso wrote:

| Re this bug:
| 
|     https://savannah.gnu.org/bugs/?34936
| 
| It looks to me like it boils down to keeping Matlab compatibility and
| making a patch object per dot. If we do that, then we really ought to
| figure out a way to optimise __go_patch__ for each possible backend,
| because it's a tremendous slowdown to do this for as little as 1000
| points.
| 
| Looks like previous attempts to not create a patch object per point
| have introduced complexity and errors into the function.
| 
| Joanna also noticed that my proposed fix only works on fltk, not on gnuplot.

I'm pretty sure there will be Matlab code out there that people would like
to run in Octave that depends on having one object per marker, so I
think we should keep compatibility.

If performance of scatter can't be improved sufficiently for the cases
you need (many markers, but only a few different types of of them, and
for which it is OK for the attributes of all the markers of a given
type to change together?) then it would probably be best to offer that
as an option or to create a separate scatter plot function that does
what you want.

jwe