`name()` should use weights from babynames dataset

[hadley]

name() currently generates mostly unusual names because it doesn't do any weighting. It would be nice if the default would use frequency weights so that you're more likely to get common names.