Switch to q0 instead of profile likelihood ratio for discovery significance via toys

TRExFitter Version and Commit Hash

master (currently 63481eaf)

Description

We use RooStats::ProfileLikelihoodTestStat for toy-based discovery significance. This looks to be only a profile likelihood ratio https://github.com/root-project/root/blob/71b931d178bec1f2e6e6a8fa0f1815a3f91dfb17/roofit/roostats/src/ProfileLikelihoodTestStat.cxx#L243, notably without a factor of 2 included. To get from this to q0 we need to multiply in a factor of 2.

Toy-based studies from @gipezzul support the lack of this factor.

As the interesting quantity for people running these toys is q0, we should multiply in that factor and plot the q0 distribution instead.

Reproducer Configuration

n/a

Ideas to Resolve

multiplying in that factor 2