Per T104150: Investigation: Make sure client change handling scales: We should make the client change handling only handle a certain amount of changes (and pages) at once (per job run). If there are more changes to handle, another job needs to be added to queue.
This causes problem were the jobs time out, run to large queries and try to purge to many pages at once.
This mostly concerns ChangeHandler::handleChange which does getAffectedUsagesByPage and then tries to handle all of these at once. In order to address this T146838: Allow seeking in UsageLookup::getPagesUsing results needs to be resolved and afterwards ChangeHandler::handleChange needs to make use of that functionality and only do part of the updates at once.
Note: Fixing this might increase the amount of updates applied to a certain wiki by a huge amount, which could in turn negatively affect other parts of the infrastructure.