documentation: add brief description of general features of models

I think we should add a new section with a brief descriptions features that are the same across models.

The case I have in mind is the description of the new nan/missing value handling.
It's described in the docstrings (I guess), but I would like to add a section BEWARE to it.
The advantages of cleaning first: You need to do it only once, if the same model is estimated repeatedly (no repeated checking and array copying). When we run regression with different number of exog variables/columns, then the number of observations can differ after removing missing observations in each exog. This is good in some cases, but not when we want to compare (nested) models.

other features ?

• pickle for prediction :?
• constant handling : cleaned up in 0.5
• t versus normal distribution in tests/postestimation inference

I don't remember if or where those are already described.

Review the existing model docstrings and documentation for nan/missing-value handling, constant handling, pickle-based prediction, and t versus normal distributions in tests and postestimation inference. Done means adding a clearly scoped general-features section, including the requested missing-value warning and any confirmed cross-model features.