GitLab

As well as being generally a good idea, os.access('foo') seems to return
true if foo.exe exists on Windows.

I'm not entirely sure of this bang-kind stuff, but in so far as it works
at all, this test is fine.  I guess I should look at the bang-kind stuff
again, as soon as the FC branch becomes the HEAD

The artificial loop wasn't running for long enough for a context
switch to happen, so the finalizer wasn't getting to run early enough.

These error messages (to do with failure in higher-rank situations)
are different, but probably no worse than before.

threadDelay is not interruptible on Windows with the threaded RTS.
Work around it in conc014, conc015 and conc017, and add a new test for
this specific failure, and mark it as an expected failure for the
relevant cases.

This means the testsuite works by default even if you have Python 2.2,
but will fail with Python 2.2 if you set THREADS.

I had to pull the global classes and instances out into their own module
as there was a catch-22: testlib needed to know if threading was enabled,
but we don't know that until we have gone through the argument, but going
through the arguments required changing things like config in testlib.