Threading just means running the iterations of a loop asynchronously rather than in order. You can thread anything, including the same analysis of different data or different analyses of the same data. As I said though, I don’t think threading is going to help you much since your jobs are computational. Multiprocessing might help, but again, PyMC3 already does multiprocessing under the hood.
Based on what you wrote here, I’ve no clue what you are trying to accomplish.