GroupBy not extensible

[Jolanrensen]

Kandy requires their own implementation of the GroupBy interface, however, thanks to:

public fun <T, G> GroupBy<T, G>.toDataFrame(groupedColumnName: String? = null): DataFrame<T> =
    if (groupedColumnName == null || groupedColumnName == groups.name()) {
        internal().df
    } else {
        internal().df.rename(groups).into(groupedColumnName)
    }

internal fun <T, G> GroupBy<T, G>.internal(): GroupByImpl<T, G> = this as GroupByImpl<T, G>

which assumes every GroupBy is GroupByImpl, this does not work.

As far as I can see, internal() is only used to retrieve the underlying df, so it should be easy enough to open this up