In surveys are EXTREME outliers considered or just thrown out when finding average?

Sometimes, an extreme outlier is thrown out, especially if it's due to human/experimental error. However, throwing out data is usually discouraged in the scientific community, so most researchers either discuss the outlier (and possible causes) in a publication or try to work around it.

Depending on the type of analysis and the spread of the data, sometimes the median is taken as the "average" instead of the mean. That's one of a few ways that people try and deal with an outlier.

As you might guess from your answers, it depends on the researcher and it depends on the data set. First and foremost, outliers are always reported. That's a matter of honesty. If you choose to discard it, though, there needs to be some discussion of why (the instruments were out of calibration that day, human or systemic error, something along those lines).

INCLUDED! If you're throwing anything out, you will want to have a good reason to do so. If the result is skewed, look for alternate measures of average such as mode or median.