One factor will be when the schools break up. End of June for Scotland, about now for rest of UK.
Note, I'm not convinced that this will be entirely due to cases dropping due to relatively less mixing between children.
Rather, children have been subjected to twice weekly testing, unlike the majority of the country, and so will identify a large %age of asymptomatic infections that would go unnoticed (and untested) in older age groups.
I anticipate a similar cliff-edge decline in positive case numbers for the rest-of-UK for this week. The question then is whether we see an uptick in the data (as the child testing results drop out of the calculation) or whether it will continue to fall (ie, this summer's infectious wave has run its course).
The authorities will, of course, declare the decline in cases as the success of the delay in allowing the relaxation of lockdown.
In addition, anyone with children will know that they worked out 'a while ago' that orange juice and coke (amongst others) could be used to get a positive result, and thus get some time off school (particularly at 'end of summer term' exam season). Sure, this will be a relatively low number of children (most are well behaved), but possibly enough to skew the 'case numbers' presented in the graph.
I'd note that the Zoe data doesn't suffer this problem (although it has other issues due to the self-selection of participants), and is probably more useful as an indication of disease progression in different areas of our Union.