As a user, I don't want a WDQS deployment to accidentally break things.
Right now, when WDQS is deployed, only service wdqs-blazegraph is restarted. This is not very useful, because:
- If we deploy updater, we need to restart wdqs-updater
- If we deploy Blazegraph, we need to restart wdqs-blazegraph and wdqs-categories
- If we only deploy GUI, we don't really need to restart either
However, since scap seems to use new checkout for each deploy, maybe in fact we need to restart all three every time, since we don't want them running off old files. Not sure yet which model would make the most sense - if we restart both, they need to be restarted one after another, with delay, otherwise Updater may fail.