Certain households see changes in head_age, not reflected in age

[prrathi]

For example, these are the instances of hh_id of 2029172029001.0 from the data in psid_lifetime_income.pkl outputted by psid_data_setup.py.

[MaxGhenis]

Thanks @prrathi. This looks to be due to the logic of

1. Taking min(head_age) per hh_id, then incrementing it according to the year (code); in this case, that's 18
2. Removing records with age < 20 (code); i.e. the first year this hh_id showed up was probably 1975, but it was dropped.

More fundamentally, this shows that the household head for a given household can change, maybe due to death or divorce. @jdebacker, how should these be handled?