I'm kicking the first part of this issue to development to see if they get something. However, the 1st approach mostly just maps an IV curve from the circuit to the lookup table version of the mosfet. You can do this manually using grabit (look on file exchange) to get the IV curves and put them into the model, and then make sure you get the correct capacitances from the data sheet and you'll have a working model without needing to work off the spice mode.
the second model isn't working because it needs to have an equivalent thermal electrical network on the tj node. At least, that is most likely the issue. I'm not sure what the id node is for (I think it's just for instrumentation?) and may or may not need to be hooked up.
