Allow for immigrants to have a different skill distribution than native population

[jdebacker]

If we let omega vary by T, S, and J we can track the fraction of the population of each age and skill group at each point in time.

With that, we can have a lambdas for immigrants that differs from natives. And the new TxSxJ omega will evolve over time as immigrants enter/leave the population.

One difficulty with this is finding reasonable empirical studies on skill levels of immigrants by age, but some approximation can be make based on data with the skill makeup of immigrants overall.