use pg-ctl stop to cleanup before killing the process

[DShau22]

Proposed fix for this issue

Ran make test to confirm that all existing tests still pass, and make check for formatting

Docs on pg-ctl for reference

[davvid]

one last request might be to rebase and squash these two commit together so that there's only one remaining, but that's mostly aesthetic. We'll let @ugtar do the final merge, tho.

[DShau22]

Just squashed and force pushed!

[ugtar]

We need to be able to override pg_ctl in the same way that we do postgres or initdb, in the case the user is using an alternative postgres install that is not in the PATH. This needs a default, and an argument (plus said default) in the constructor. See for example DEFAULT_POSTGRES or DEFAULT_INITDB

[ugtar]

I think since run_cmd with level 2 can potentially produce stdout and stderr, instead of .wait() we should call .communicate(). Also, since this command can be caught by the pg process or hang, we should have a timeout at which point we give up and call kill() anyway. Maybe 10 seconds?
The python docs have an example of how to do this properly, catching the TimeoutExpired exception https://docs.python.org/3/library/subprocess.html#subprocess.Popen.communicate

[ugtar]

thanks @DShau22, sorry I took so long to get to this, I've been swamped at dayjob