Improve aggregation across entire dataframe

[timsaucer]

From this comment:

I wonder if it would feel more pythonic to accept None in addition to the empty list if the description is No grouping

Originally posted by @ntjohnson1 in #1498 (comment)

This is in reference to doing aggregation like

df.aggregate([], [f.min(col("a")).alias("min_a")])

The suggestion being that maybe it would be nice to take

df.aggregate(None, [f.min(col("a")).alias("min_a")])