pct_combing_weight scaling issue

[northroj]

In kernel.py

pct_combing_weight sets pc_factor with the following code:

    # Teeth distance
    td = W / M

    # Update population control factor
    mcdc["technique"]["pc_factor"] *= td  # This may be incorrect

This also applies to pct_splitting_roulette_weight

pc_factor is used to rescale weight windows at each time step

    # Population control factor
    w_target *= mcdc["technique"]["pc_factor"]

Which works for uniform combing but may be unhealthy for the weight based PCTs when combined with weight windows.