Temporary fix: Spec test timeouts don't work on Windows

See #63 (moved)

Temporary workaround for test timeouts: when the test's processes fails with a pickle error, rerun the test without timeout. This rerun is done as a subtest, so it displays that the timeout was disabled.

Note: this may cause IDEs to show an incorrect amount of tests, as subtests can be seen as a new test case. The python testrunner still shows the correct amount.