Pyspark Series2Series UDF with Array of Structs as input and Struct as output

Question

How can I construct a UDF in spark which has nested (struct) input and output values?

NOTICE: I am aware of certain limitations of older versions of Arro