Deassert start probe after test is done

[martijnbastiaan]

We currently deassert a HITL start probe before starting a test (and subsequently assert it again). We do this such that the probe stays asserted after finishing a test, which allows us to probe CPUs over JTAG during the postprocessing step. This is an enormous hack though, and should be resolved by #502. After merging #502 we should make sure to deassert start probes after finishing a test again.

TODO

• Link relevant code after merging Move test control logic to Haskell #536