[merge][rfc] remove python -O from make check

[Andrew Bennetts]

Martin Pool wrote:
[...]
> 
> I wonder if we should also change them to run selftest -1 (stop on
> first error)?  I'd prefer to get information about one error sooner
> than about many errors later.  It's almost always something that I can
> reproduce locally.

That sounds reasonable to me.

After all, if a test failure from PQM is mysterious, and you think having more
than one test failure to look at would be helpful, you could always submit a
request for your failing branch + a commit to remove the -1 option.  But I doubt
anyone will ever need to go to such lengths.

-Andrew.