Some cleanup of pg_stat_statements tests

Hi all,

While looking at this area of the code, I have noticed that the new
"squashing" test is not consistent with the rest of the test suite,
re-creating the extension for nothing.

The structure of the test is to have the "cleanup" phase second to
last, so as there is no need to create multiple times the extensions.

In passing, I'd like to suggest that REGRESS in the Makefile uses one
item per line, to minimize diffs when we add new stuff.

Any thoughts or objections about the attached?
Thanks,
--
Michael