Run out of memory allocating stoichiometry matrix

[rwest]

This seems to be the limiting factor on how large a model you can build

[rwest]

Stoichiometry matrix is now sparse. This should help memory problems.

Had to replace a few calls to numpy.dot, but I think I got it working again.
Some info on sparse matrices in numpy can be found at
http://www.scipy.org/SciPy_Tutorial

Closed by 401aff2 (hopefully this linking to the issue tracker works!)

[rwest]

Reverted by e525b08 because creating the sparse lilmatrix was very very slow.
We need a better way to solve this

[rwest]

I think this is solved again by Josh, but I can't remember when.