Post-processing stages are very slow

The steps in data analysis after the sampling completes are extremely slow as they don't JIT the function calls.

We should see if there is a way to speed some of these things up. Jitting the individual model functions might help with this.