matplotlib priorities

suggestions:
(1) a color bar / scale for the pseudo color / image plots.
A graphic mapping of values to colors is pretty standard in such displays.

(2) Although map projections are listed last in the 'Plot types' category, it
would be good to consider how these will be implemented right now.
Map projections are just a particular application of a user coordinate ->display coordinate
transformation, and such a transformation capability needs to be at the core of all the plots.
They are more handy than one might think.
If such a transformation hook is available then the map projections are straightforward using
PROJ4 ( http://www.remotesensing.org/proj/ ) code, for example.
Actually, the trickiest part of coordinate transformations is positioning the axis labels in
  a 'nice' way.

Astronomers use projections too - the whole sky COBE displays appear to be in a Hammer or Mollweide projection.

Jim