Get more information about failing CI runs

[linusha]

Currently, when tests fail in CI and we cannot reproduce the failure locally, we are more or less unable to gather any significant information about what the problem is.
There are two aspects to this:

1. Error messages from within lively or the output of failing tests should be retrieved. This can be achieved with simple changes to the test runner, the testing subserver and the CI scripts.
2. Sometimes, for example in the cases of Some lively.morphic/text tests regularly fail in CI #1524, this information would not suffice. What is needed there would be to capture the logs of the console of the chrome instance running the headless tests. More research is needed to find a suitable solution for this problem.