display hooks for matplotlib

[stevengj]

Requires a bit of surgery to get matplotlib to use our new display functionality in order to get inline plots.

A model here is provided by pylab/backend_inline.py and the corresponding code in core/pylabtools.py from IPython, as pointed out by @fperez

[stevengj]

As a prerequisite, I just added functionality to PyCall so that it can pass Julia IO objects as native Python IO objects (along with the patch JuliaLang/julia#3878).

[stevengj]

Until DataDisplay is merged into Base, at which point this can be merged directly into PyCall, IJulia now includes a wrapper module IPylab. Doing using IPylab imports Pylab as plt, and adds the hooks to display figures inline.

I haven't written a proper Pylab backend yet. So far I've only implemented the ability to display Figure objects. So, if you do:

plt.plot(.....)
plt.gcf()

the gcf function will return a figure object that is displayed graphically inline.