Add Pair iterator sink to IndexedTables

[davidanthoff]

The IndexedTable sink should accept iterator of Pairs of NamedTuples, and then use the columns from the key as index columns and the columns from the value as data columns. This would enable e.g. this query:

@from i in df begin
@select {i.a, i.b} => {i.c, i.d}
@collect IndexedTable
end

Here the index columns would be a and b, and the data columns c and d.