I should have updated this in commits 52a10224e and follow-ons,
but I missed it because it's not run by default, and none of the
buildfarm runs it either.  Maybe we should try to improve that
situation.

Discussion: https://postgr.es/m/CAH2-Wz=j9SRW=s5BV4-3k+=tr4N3A03in+gTuVA09vNF+-iHjA@mail.gmail.com

A couple of test cases had connect_timeout=14, a value that seems
to have been plucked from a hat.  While it's more than sufficient
for normal cases, slow/overloaded buildfarm machines can get a timeout
failure here, as per recent report from "sungazer".  Increase to 180
seconds, which is in line with our typical timeouts elsewhere in
the regression tests.

Back-patch to 9.6; the code looks different in 9.5, and this doesn't
seem to be quite worth the effort to adapt to that.

Report: https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=sungazer&dt=2020-08-04%2007%3A12%3A22

The documentation used "nonexistent" exclusively, and the source tree
used it three times as often as "nonexistant".

To ensure that "make installcheck" can be used safely against an existing
installation, we need to be careful about what global object names
(database, role, and tablespace names) we use; otherwise we might
accidentally clobber important objects.  There's been a weak consensus that
test databases should have names including "regression", and that test role
names should start with "regress_", but we didn't have any particular rule
about tablespace names; and neither of the other rules was followed with
any consistency either.

This commit moves us a long way towards having a hard-and-fast rule that
regression test databases must have names including "regression", and that
test role and tablespace names must start with "regress_".  It's not
completely there because I did not touch some test cases in rolenames.sql
that test creation of special role names like "session_user".  That will
require some rethinking of exactly what we want to test, whereas the intent
of this patch is just to hit all the cases in which the needed renamings
are cosmetic.

There is no enforcement mechanism in this patch either, but if we don't
add one we can expect that the tests will soon be violating the convention
again.  Again, that's not such a cosmetic change and it will require
discussion.  (But I did use a quick-hack enforcement patch to find these
cases.)

Discussion: <16638.1468620817@sss.pgh.pa.us>

fix segfault on non-glibc systems.

create expected file with correct port number

Made this option mark the .c files, so the environment variable is no longer needed.
Created a special MinGW file with the special error message.
Do not print port into log file when running regression tests.