Fix some typos

[GreatArcStudios]

Hey Scott,

These are some of the typos I mentioned over coffee the other day...the one that really caught my eye was "endog" for $\mathbf{X}$'s. Seems like it should be exog if we are talking about the term favoured by the economist types (see statsmodels here).

Also under The Linear Regression Model [20 minutes], I think it'd be more clear if conditional expectation was explicitly mentioned somewhere since it's good to understand that LR is a conditional means model. Cool stuff!