use correct beta_u in shallow water linear solvers

[jshipton]

Sorry @Witt-D I didn't notice this in your PR.

[tommbendall]

This puts things in line with the rest of the code ... but the awkward thought I've had is that maybe we don't want the beta values to be constant at all?

In theory we could update dt, but if we did then we would want the beta values to be updated too. At the moment they won't be, so do we actually want:

beta = self.dt*Constant(tau)

instead of

beta_ = self.dt*tau
beta = Constant(beta_)