Specifying the test number manually is tedious and prone to errors (as
recently proven). Since we have all the necessary data to work out the
test number, let's do it automagically.
The logic to query test state was rather complex. I don't quite grok the point
of ret=$((ret+1))… But afaics, the precise result was always ignored by the
caller anyway.
Building custom images for each test takes a lot of time.
Build the default one, and if the test needs incompatible changes
just copy it and extend it instead.
When tests are executed serially (the default), it seems better to launch
the fairly generic test that runs the unittests early in the sequence.
Right now the tests are ordered based on when they were written, but
this doesn't make much sense.
