Ola tried to change the matching scale consistently in the ME and TMD. When doing this, the bump in the ratio does not disappear. It is speculated that this might be caused by an issue with the matching not being perfectly correct (double-counting some phase space regions) or some more fundamental parameter (the value of alphas). In the first case, Herwig 6 is expected to not have the bump, which could technically be tested.
If an issue with the matching is found, this would probably embark us into a longer study. For our current project, we agreed that we could simply bring it to people's attention as something that affects the extracted value of qs.