Add a timeout the NuxeoDrive.GetChangeSummary call to prevent such huge times as reported in the attached CSV file. The interesting column in "Event Label" which is the number of seconds the call take to complete.
Values are sorted in descendant order (longer -> shorter calls). Some are really huge and should be discared. I am not sure how to get such values though. Maybe the computer that goes sleeping?
In any cases, as we are polling for changes on a regular basis (defined by the delay option, defaults to 30 seconds), it makes sense to limit with a timeout the time that call could take.