When I try to us axes.imshow in matplotlib with a grid of known size and shape,
and then later read the png data, the image's data changes (is shifted) compared to the
input gridded data.

There may be a flag or option that I'm just not aware of, so I would love any help to prevent this shift.

Please see the gist included below for a self contained example.

To help us understand and resolve your issue please check that you have provided
the information below.

• [x ] Matplotlib version, Python version and Platform (Windows, OSX, Linux ...)

 $ python --version
Python 3.5.2
vagrant@vmsnowoffpiste ~/projects/cartopy-why (master *=)
 $ conda list | grep matplotlib
matplotlib                1.5.3               np111py35_3    conda-forge/channel/main

I also ran my test code on macosx+conda with the same results.

• [X ] How did you install Matplotlib and Python (pip, anaconda, from source ...)
  via conda
• [ X] If possible please supply a Short, Self Contained, Correct, Example
  that demonstrates the issue i.e a small piece of code which reproduces the issue
  and can be run with out any other (or as few as possible) external dependencies.
  Please see gist here (has small code and data demonstrating the problem):
  https://gist.github.com/flamingbear/8f3e081c31fbb428fe60afe71729d5ec
• [ X ] If this is an image generation bug attach a screenshot demonstrating the issue.
  Easier to generate the problem image with the code from the gist above, but here are the output images I'm getting.
  flickering gif: http://imgur.com/i8Tf16X
  imshow's png: http://imgur.com/syjZmQ4
  scipy's png: http://imgur.com/6rLIxA4