We weren't testing anything involving EPQ on UPDATEs that move tuples
into different partitions.  Depending on the implementation,
it might be that these cases aren't actually very interesting ...
but given our thin coverage of EPQ in general, I think it's a good
idea to have a test case.

Amit Langote, minor tweak by me

Discussion: https://postgr.es/m/7889df35-ad1a-691a-00e3-4d4b18f364e3@lab.ntt.co.jp