I bought and read a book about multiple regression analysis (least squares) and it contains tests for each assumptions and not just linear models, but non-linear ones, too. It also contains dummy variables, etc.

If I knew the techniques in this book inside out and understood them and knew how to apply that knowledge, would that be all I have to know about multiple regression analysis (other than logistic regression)?

For example with time series data one should look at the plot and make sure to check for auto-correlation. However, that should probably be done with every data set. However, I guess it must be more complicated than that :-).

So say I learn these techniques about multiple linear regression and completely understand all of them (why the assumptions are made, when they might be hurt, how to test if they are hurt) is there anything else (theory-wise) I would have to know?

I assume, there's still more one would have to know to be really good at regression analysis. Does this mean the difference between somebody who only has the basics down (but doesn't hurt any assumptions) and somebody who's an expert at it, is that the expert will be able to create a model that fits the data better than that of the intermediate student?