`nsmallest` doesn't work if the `dtype` is `object` but `sort_values()` does

[baziotis]

Example:

df = pd.read_csv('https://web.stanford.edu/class/archive/cs/cs109/cs109.1166/stuff/titanic.csv')
df['Name'].nsmallest(n=5)

Add another precondition to SortHead. It makes sense to solve issue #3 first.