Add rowwise function

[pdeffebach]

Iterates through all rows in a dataframe and makes an array of length nrow(df).

[pdeffebach]

usage is t = rowwise(mean, df)

If you approve the concept, I will explore performance more deeply. mapslices(f, x, 2) is super fast, but i'm not sure how to get close to that performance with dataframes, which was is better optimized for column operations.

[nalimilan]

Now that iteration over DataFrameRow yields values, this can be written as [mean(r) for r in eachrow(df)], so I don't think we need this. We could also get rid of colwise. See #1595.