Optimal lattice rescaling for EV curves

[lukeamyers]

Uniform lattice scaling for the next volume leads to an inefficient guess. So the calculation of the energy at the next point takes longer than it should. It should be possible to consider how each lattice parameter has change from the previous point on the EV curve and scale accordingly. This is a minor issue, but could be helpful for materials that have significant stress-strain anisotropy.

For example, graphite will stretch and compress more along the normal vector of the graphene sheets, so the POSCAR should be scaled accordingly between each volume calculation for a better initial guess.

Just putting this here for future consideration.