Add another staker test to make sure the full coverage of fraud proof code

[ImJeremyHe]

I encountered a staker test error natively but can not reproduce it now. (Maybe some environment issues?)
In the current e2e test, good staker always acts first and if I understand correctly, the challenge should always be ended up by the osp proof provided by the bad staker.
Although we have an OSP test already to check the L1 arbitrator, it may not be enough.
It should be better to write one more staker test.