Much better thermodynamics now, they are pretty much done for this stage.
Instead of just throwing heat in directly from plumes, now every single thing is done via actual heat flow based on temperature gradients, so it gets stronger the bigger the gradient. This strongly encourages equilibria and stability.
In this image, the three vertical strips are cutaways for debugging purposes. Left side is top surface layer, middle is a middle layer of the crust, right is the layer nearest the mantle.
The mantle in general has a temperature, and so does the atmosphere, as well as thermal qualities like conductance and specific heat. In the vicinity of plumes, the mantle temperature goes up locally (and it's hard to see, but it doesn't add blindly, but acts sort of convection cell-shaped, only primitively though). So everything cools to air on top and heats to the mantle beneath, and settles into equilibrium states.
Currently, on a 300x300x5-thick world, the calculating flow between every goxel takes about 1/3 of a second (representing 1 million years or so). I had to do a bit of optimizing to make it sane, but done for now. Later if it's an issue, I can look into threading temperature separately or checking small chunks and not updating them if they are staying static, etc.
Less glamorous than the last image, but also wouldn't lead to giant melted holes in the crust unlike that one =P
(at equilibrium with non-moving plates, after 75 or so steps)
Later, as well, the temperature will be more interesting. When the thickness of the crust begins to vary, the proximal mantle will be hotter further down than for the thinner crust. And new ocean plates will be a bit warmer when just formed, and the local mantle temp will lower temporarily if cold rock is subducting into it nearby, and different materials with different thermal properties mix in, etc.
Also just for fun, one of the more beautiful buggy outputs I had while trying to get this to work properly. IIRC, it thought that the plates were vibrating and gliding, or something like that, when they weren't supposed to be (this was before the equilibrium, so red just stayed red):
It also gives you an idea of how much is actually going on under the hood beyond what the nice little symmetrical images suggest (a side effect of plates not moving yet, etc.)