Test coverage for calculate_upper_limits() in core/unblinding.py

[mlincett]

It seems the core.unblinding is not properly covered by testing.

In fact, PR #242 at the time of writing should break unblinding but automated testing is successful.

It would be unfortunate to have an analysis approved for unblinding and find out only at the end that the tagged flarestack version is broken.

[mlincett]

I may be partially wrong in the sense that I expect unblinding to be tested in test_full_analysis_chain.py. It is only calculate_upper_limits() that lacks testing.