Upgrade matplotlib version shipped on Windows

[Russell Taylor]

The change to the v2 PyQt API (#7950) has broken the import the version of matplotlib that we ship. Upgrade to the latest version, which has apparently resolved this issue.

Also check things on linux - on RHEL6 for example the rpm version of matplotlib is 0.99.

[Russell Taylor]

Re #8235. Remove existing version of matplotlib (32 bit).

Changeset: 0a934ab6ec43fed5abc513000f75ce26ee6604d2

[Russell Taylor]

Re #8235. Add matplotlib 1.3.1 (32 bit).

Changeset: eba82839557b7073d2ea1cc60cccb329ab6a3273

[Russell Taylor]

Re #8235. Latest matplotlib requires pyparsing (this is v2.0.1).

Changeset: f2b83171ba498c1a004640f3b7b718630a213631

[Russell Taylor]

Re #8235. Remove existing version of matplotlib (64 bit).

Changeset: cdbe8ba54e58894aa876dfded21168f562b3f8da

[Russell Taylor]

Re #8235. Add matplotlib 1.3.1 (64 bit).

Changeset: 0f2c5f9890131ee1a43a8548da8afda57f3c2baa

[Russell Taylor]

Re #8235. Add pyparsing v2.0.1

Changeset: 2fa6b8c412db8cc49a362ef926efb36c24795603

[Russell Taylor]

Ubuntu 12.04 has version 1.1.1, which should be fine.

RHEL6 has 0.99.1.1, which appears to work fine - even though a version higher than this was apparently not OK on Windows.

We don't ship matplotlib on the Mac, so there's nothing to check there.

[Russell Taylor]

Unsurprisingly, this has to be tested on windows.

The changes are only in the Third Party repositories, where we don't use branches - so there's nothing to merge and this can be tested on any build after the changes were made.

To test: Download the nightly build (or a package from one of the jenkins jobs if it's still Friday) and check that you can import matplotlib in the IPython console. Please also try the specific command which the reporting user was trying: import matplotlib.pyplot as plt. Please test both the 64 & 32 bit packages.

[Martyn Gigg]

Checked both 32- && 64-bit packages and matplotlib works correctly on both.

I tested it with

import numpy as np
import matplotlib.pyplot as plt

x = np.arange(0, 5, 0.1);
y = np.sin(x)
plt.plot(x, y)

[Stuart Campbell]

This ticket has been transferred to github issue ​9080