Currently only CSIT-1901 is sometimes causing invalid data, and that can be fixed separately. The open questions for this bug:

1. What to do when passing test is found creating invalid data? Turn it into failing test and re-export? Not straightforward as failure in Python generates message in Robot to be re-exported in Python.

2. What to do when failing test (direct or due to invalid pass data) is found creating invalid data? Have a trusted template that cannot become invalid?

In general we want to avoid not exporting anything, as that is what happens if the test has not run at all (e.g. suite setup failure or jobspec typo).

Setting low priority, as likelihood of periodic job causing invalid data outside CSIT-1901 is negligible.