Cahn-Hilliard documentation has a sign wrong

[Patrick Farrell]

In demo/pde/cahn-hilliard/common.txt, it describes the weak form of the equation for \mu (line 52) as

(\mu) - (df/dc) + (\lambda \nabla c \cdot \nabla v) = 0

where the important point is the last term is positive, the opposite sign to df/dc.
However, the cahn-hilliard.py (line 95) has

L1 = mu*v*dx - dfdc*v*dx - lmbda*dot(grad(c), grad(v))*dx

which has the last term as negative, the same sign as df/dc.

The same sign error (\lambda should be -\lambda everywhere) occurs throughout the common.txt. (I'm pretty sure the sign should be negative, comparing to http://projecteuclid.org/euclid.cms/1214949925)