Fix Ta bug

[finsberg]

Fix bug in strong coupling.

Previously we have been solving

$$ 0 = \nabla \cdot P = \nabla \cdot \frac{d(\psi + Ta(\lambda)(I_{4f} - I))}{dF} $$

Now we are instead solving

$$ 0 = \nabla \cdot P = \nabla \cdot ( \frac{d \psi}{dF} + Ta(\lambda)\frac{d (I_{4f} - I)}{dF}) $$

Adopted from https://github.com/ComputationalPhysiology/simcardems2