Best approaches for grouped median

I revisit an old post and explores approaches to calculating the median across partitioned sets.